├── LICENSE
├── README.md
└── scripts
├── external_data
├── Randolph2021
│ ├── mock_pbmc_processing.R
│ └── monocytes_trajectory.R
└── Smillie2019
│ └── singlecell_analysis.R
├── finemap
├── 01_calc_LD.sh
├── 01_normalize_blueprint.R
├── 02_merge_pseudobulk_blueprint_snps.sh
├── 02_prep_finemap_files.R
├── 03_merge_pseudobulk_blueprint_resid.R
└── 03_run_finemap.sh
├── post_analysis
├── calc_singlecell_betas.R
├── cluster_eGenes.R
├── enrich_atac.R
├── enrich_impact.R
├── enrich_tss.R
├── run_coloc.R
└── run_homer.sh
├── pseudobulk
├── 01_make_pseudobulk_exprs.R
├── 02_normalize.R
├── 03_make_BED.sh
├── 04_fastqtl_nominal.sh
└── 05_fastqtl_permute.sh
└── singlecell
├── linear_nostate.R
├── linear_nostate_permute.R
├── linear_univariate.R
├── linear_univariate_permute.R
├── linear_univariate_simulDE.R
├── nb_univariate.R
├── poisson_multivariate.R
├── poisson_nostate.R
├── poisson_nostate_permute.R
├── poisson_univariate.R
├── poisson_univariate_condition.R
├── poisson_univariate_permute.R
├── poisson_univariate_simulDE.R
└── poisson_univariate_simulnull.R
/LICENSE:
--------------------------------------------------------------------------------
1 | GNU GENERAL PUBLIC LICENSE
2 | Version 3, 29 June 2007
3 |
4 | Copyright (C) 2007 Free Software Foundation, Inc.
5 | Everyone is permitted to copy and distribute verbatim copies
6 | of this license document, but changing it is not allowed.
7 |
8 | Preamble
9 |
10 | The GNU General Public License is a free, copyleft license for
11 | software and other kinds of works.
12 |
13 | The licenses for most software and other practical works are designed
14 | to take away your freedom to share and change the works. By contrast,
15 | the GNU General Public License is intended to guarantee your freedom to
16 | share and change all versions of a program--to make sure it remains free
17 | software for all its users. We, the Free Software Foundation, use the
18 | GNU General Public License for most of our software; it applies also to
19 | any other work released this way by its authors. You can apply it to
20 | your programs, too.
21 |
22 | When we speak of free software, we are referring to freedom, not
23 | price. Our General Public Licenses are designed to make sure that you
24 | have the freedom to distribute copies of free software (and charge for
25 | them if you wish), that you receive source code or can get it if you
26 | want it, that you can change the software or use pieces of it in new
27 | free programs, and that you know you can do these things.
28 |
29 | To protect your rights, we need to prevent others from denying you
30 | these rights or asking you to surrender the rights. Therefore, you have
31 | certain responsibilities if you distribute copies of the software, or if
32 | you modify it: responsibilities to respect the freedom of others.
33 |
34 | For example, if you distribute copies of such a program, whether
35 | gratis or for a fee, you must pass on to the recipients the same
36 | freedoms that you received. You must make sure that they, too, receive
37 | or can get the source code. And you must show them these terms so they
38 | know their rights.
39 |
40 | Developers that use the GNU GPL protect your rights with two steps:
41 | (1) assert copyright on the software, and (2) offer you this License
42 | giving you legal permission to copy, distribute and/or modify it.
43 |
44 | For the developers' and authors' protection, the GPL clearly explains
45 | that there is no warranty for this free software. For both users' and
46 | authors' sake, the GPL requires that modified versions be marked as
47 | changed, so that their problems will not be attributed erroneously to
48 | authors of previous versions.
49 |
50 | Some devices are designed to deny users access to install or run
51 | modified versions of the software inside them, although the manufacturer
52 | can do so. This is fundamentally incompatible with the aim of
53 | protecting users' freedom to change the software. The systematic
54 | pattern of such abuse occurs in the area of products for individuals to
55 | use, which is precisely where it is most unacceptable. Therefore, we
56 | have designed this version of the GPL to prohibit the practice for those
57 | products. If such problems arise substantially in other domains, we
58 | stand ready to extend this provision to those domains in future versions
59 | of the GPL, as needed to protect the freedom of users.
60 |
61 | Finally, every program is threatened constantly by software patents.
62 | States should not allow patents to restrict development and use of
63 | software on general-purpose computers, but in those that do, we wish to
64 | avoid the special danger that patents applied to a free program could
65 | make it effectively proprietary. To prevent this, the GPL assures that
66 | patents cannot be used to render the program non-free.
67 |
68 | The precise terms and conditions for copying, distribution and
69 | modification follow.
70 |
71 | TERMS AND CONDITIONS
72 |
73 | 0. Definitions.
74 |
75 | "This License" refers to version 3 of the GNU General Public License.
76 |
77 | "Copyright" also means copyright-like laws that apply to other kinds of
78 | works, such as semiconductor masks.
79 |
80 | "The Program" refers to any copyrightable work licensed under this
81 | License. Each licensee is addressed as "you". "Licensees" and
82 | "recipients" may be individuals or organizations.
83 |
84 | To "modify" a work means to copy from or adapt all or part of the work
85 | in a fashion requiring copyright permission, other than the making of an
86 | exact copy. The resulting work is called a "modified version" of the
87 | earlier work or a work "based on" the earlier work.
88 |
89 | A "covered work" means either the unmodified Program or a work based
90 | on the Program.
91 |
92 | To "propagate" a work means to do anything with it that, without
93 | permission, would make you directly or secondarily liable for
94 | infringement under applicable copyright law, except executing it on a
95 | computer or modifying a private copy. Propagation includes copying,
96 | distribution (with or without modification), making available to the
97 | public, and in some countries other activities as well.
98 |
99 | To "convey" a work means any kind of propagation that enables other
100 | parties to make or receive copies. Mere interaction with a user through
101 | a computer network, with no transfer of a copy, is not conveying.
102 |
103 | An interactive user interface displays "Appropriate Legal Notices"
104 | to the extent that it includes a convenient and prominently visible
105 | feature that (1) displays an appropriate copyright notice, and (2)
106 | tells the user that there is no warranty for the work (except to the
107 | extent that warranties are provided), that licensees may convey the
108 | work under this License, and how to view a copy of this License. If
109 | the interface presents a list of user commands or options, such as a
110 | menu, a prominent item in the list meets this criterion.
111 |
112 | 1. Source Code.
113 |
114 | The "source code" for a work means the preferred form of the work
115 | for making modifications to it. "Object code" means any non-source
116 | form of a work.
117 |
118 | A "Standard Interface" means an interface that either is an official
119 | standard defined by a recognized standards body, or, in the case of
120 | interfaces specified for a particular programming language, one that
121 | is widely used among developers working in that language.
122 |
123 | The "System Libraries" of an executable work include anything, other
124 | than the work as a whole, that (a) is included in the normal form of
125 | packaging a Major Component, but which is not part of that Major
126 | Component, and (b) serves only to enable use of the work with that
127 | Major Component, or to implement a Standard Interface for which an
128 | implementation is available to the public in source code form. A
129 | "Major Component", in this context, means a major essential component
130 | (kernel, window system, and so on) of the specific operating system
131 | (if any) on which the executable work runs, or a compiler used to
132 | produce the work, or an object code interpreter used to run it.
133 |
134 | The "Corresponding Source" for a work in object code form means all
135 | the source code needed to generate, install, and (for an executable
136 | work) run the object code and to modify the work, including scripts to
137 | control those activities. However, it does not include the work's
138 | System Libraries, or general-purpose tools or generally available free
139 | programs which are used unmodified in performing those activities but
140 | which are not part of the work. For example, Corresponding Source
141 | includes interface definition files associated with source files for
142 | the work, and the source code for shared libraries and dynamically
143 | linked subprograms that the work is specifically designed to require,
144 | such as by intimate data communication or control flow between those
145 | subprograms and other parts of the work.
146 |
147 | The Corresponding Source need not include anything that users
148 | can regenerate automatically from other parts of the Corresponding
149 | Source.
150 |
151 | The Corresponding Source for a work in source code form is that
152 | same work.
153 |
154 | 2. Basic Permissions.
155 |
156 | All rights granted under this License are granted for the term of
157 | copyright on the Program, and are irrevocable provided the stated
158 | conditions are met. This License explicitly affirms your unlimited
159 | permission to run the unmodified Program. The output from running a
160 | covered work is covered by this License only if the output, given its
161 | content, constitutes a covered work. This License acknowledges your
162 | rights of fair use or other equivalent, as provided by copyright law.
163 |
164 | You may make, run and propagate covered works that you do not
165 | convey, without conditions so long as your license otherwise remains
166 | in force. You may convey covered works to others for the sole purpose
167 | of having them make modifications exclusively for you, or provide you
168 | with facilities for running those works, provided that you comply with
169 | the terms of this License in conveying all material for which you do
170 | not control copyright. Those thus making or running the covered works
171 | for you must do so exclusively on your behalf, under your direction
172 | and control, on terms that prohibit them from making any copies of
173 | your copyrighted material outside their relationship with you.
174 |
175 | Conveying under any other circumstances is permitted solely under
176 | the conditions stated below. Sublicensing is not allowed; section 10
177 | makes it unnecessary.
178 |
179 | 3. Protecting Users' Legal Rights From Anti-Circumvention Law.
180 |
181 | No covered work shall be deemed part of an effective technological
182 | measure under any applicable law fulfilling obligations under article
183 | 11 of the WIPO copyright treaty adopted on 20 December 1996, or
184 | similar laws prohibiting or restricting circumvention of such
185 | measures.
186 |
187 | When you convey a covered work, you waive any legal power to forbid
188 | circumvention of technological measures to the extent such circumvention
189 | is effected by exercising rights under this License with respect to
190 | the covered work, and you disclaim any intention to limit operation or
191 | modification of the work as a means of enforcing, against the work's
192 | users, your or third parties' legal rights to forbid circumvention of
193 | technological measures.
194 |
195 | 4. Conveying Verbatim Copies.
196 |
197 | You may convey verbatim copies of the Program's source code as you
198 | receive it, in any medium, provided that you conspicuously and
199 | appropriately publish on each copy an appropriate copyright notice;
200 | keep intact all notices stating that this License and any
201 | non-permissive terms added in accord with section 7 apply to the code;
202 | keep intact all notices of the absence of any warranty; and give all
203 | recipients a copy of this License along with the Program.
204 |
205 | You may charge any price or no price for each copy that you convey,
206 | and you may offer support or warranty protection for a fee.
207 |
208 | 5. Conveying Modified Source Versions.
209 |
210 | You may convey a work based on the Program, or the modifications to
211 | produce it from the Program, in the form of source code under the
212 | terms of section 4, provided that you also meet all of these conditions:
213 |
214 | a) The work must carry prominent notices stating that you modified
215 | it, and giving a relevant date.
216 |
217 | b) The work must carry prominent notices stating that it is
218 | released under this License and any conditions added under section
219 | 7. This requirement modifies the requirement in section 4 to
220 | "keep intact all notices".
221 |
222 | c) You must license the entire work, as a whole, under this
223 | License to anyone who comes into possession of a copy. This
224 | License will therefore apply, along with any applicable section 7
225 | additional terms, to the whole of the work, and all its parts,
226 | regardless of how they are packaged. This License gives no
227 | permission to license the work in any other way, but it does not
228 | invalidate such permission if you have separately received it.
229 |
230 | d) If the work has interactive user interfaces, each must display
231 | Appropriate Legal Notices; however, if the Program has interactive
232 | interfaces that do not display Appropriate Legal Notices, your
233 | work need not make them do so.
234 |
235 | A compilation of a covered work with other separate and independent
236 | works, which are not by their nature extensions of the covered work,
237 | and which are not combined with it such as to form a larger program,
238 | in or on a volume of a storage or distribution medium, is called an
239 | "aggregate" if the compilation and its resulting copyright are not
240 | used to limit the access or legal rights of the compilation's users
241 | beyond what the individual works permit. Inclusion of a covered work
242 | in an aggregate does not cause this License to apply to the other
243 | parts of the aggregate.
244 |
245 | 6. Conveying Non-Source Forms.
246 |
247 | You may convey a covered work in object code form under the terms
248 | of sections 4 and 5, provided that you also convey the
249 | machine-readable Corresponding Source under the terms of this License,
250 | in one of these ways:
251 |
252 | a) Convey the object code in, or embodied in, a physical product
253 | (including a physical distribution medium), accompanied by the
254 | Corresponding Source fixed on a durable physical medium
255 | customarily used for software interchange.
256 |
257 | b) Convey the object code in, or embodied in, a physical product
258 | (including a physical distribution medium), accompanied by a
259 | written offer, valid for at least three years and valid for as
260 | long as you offer spare parts or customer support for that product
261 | model, to give anyone who possesses the object code either (1) a
262 | copy of the Corresponding Source for all the software in the
263 | product that is covered by this License, on a durable physical
264 | medium customarily used for software interchange, for a price no
265 | more than your reasonable cost of physically performing this
266 | conveying of source, or (2) access to copy the
267 | Corresponding Source from a network server at no charge.
268 |
269 | c) Convey individual copies of the object code with a copy of the
270 | written offer to provide the Corresponding Source. This
271 | alternative is allowed only occasionally and noncommercially, and
272 | only if you received the object code with such an offer, in accord
273 | with subsection 6b.
274 |
275 | d) Convey the object code by offering access from a designated
276 | place (gratis or for a charge), and offer equivalent access to the
277 | Corresponding Source in the same way through the same place at no
278 | further charge. You need not require recipients to copy the
279 | Corresponding Source along with the object code. If the place to
280 | copy the object code is a network server, the Corresponding Source
281 | may be on a different server (operated by you or a third party)
282 | that supports equivalent copying facilities, provided you maintain
283 | clear directions next to the object code saying where to find the
284 | Corresponding Source. Regardless of what server hosts the
285 | Corresponding Source, you remain obligated to ensure that it is
286 | available for as long as needed to satisfy these requirements.
287 |
288 | e) Convey the object code using peer-to-peer transmission, provided
289 | you inform other peers where the object code and Corresponding
290 | Source of the work are being offered to the general public at no
291 | charge under subsection 6d.
292 |
293 | A separable portion of the object code, whose source code is excluded
294 | from the Corresponding Source as a System Library, need not be
295 | included in conveying the object code work.
296 |
297 | A "User Product" is either (1) a "consumer product", which means any
298 | tangible personal property which is normally used for personal, family,
299 | or household purposes, or (2) anything designed or sold for incorporation
300 | into a dwelling. In determining whether a product is a consumer product,
301 | doubtful cases shall be resolved in favor of coverage. For a particular
302 | product received by a particular user, "normally used" refers to a
303 | typical or common use of that class of product, regardless of the status
304 | of the particular user or of the way in which the particular user
305 | actually uses, or expects or is expected to use, the product. A product
306 | is a consumer product regardless of whether the product has substantial
307 | commercial, industrial or non-consumer uses, unless such uses represent
308 | the only significant mode of use of the product.
309 |
310 | "Installation Information" for a User Product means any methods,
311 | procedures, authorization keys, or other information required to install
312 | and execute modified versions of a covered work in that User Product from
313 | a modified version of its Corresponding Source. The information must
314 | suffice to ensure that the continued functioning of the modified object
315 | code is in no case prevented or interfered with solely because
316 | modification has been made.
317 |
318 | If you convey an object code work under this section in, or with, or
319 | specifically for use in, a User Product, and the conveying occurs as
320 | part of a transaction in which the right of possession and use of the
321 | User Product is transferred to the recipient in perpetuity or for a
322 | fixed term (regardless of how the transaction is characterized), the
323 | Corresponding Source conveyed under this section must be accompanied
324 | by the Installation Information. But this requirement does not apply
325 | if neither you nor any third party retains the ability to install
326 | modified object code on the User Product (for example, the work has
327 | been installed in ROM).
328 |
329 | The requirement to provide Installation Information does not include a
330 | requirement to continue to provide support service, warranty, or updates
331 | for a work that has been modified or installed by the recipient, or for
332 | the User Product in which it has been modified or installed. Access to a
333 | network may be denied when the modification itself materially and
334 | adversely affects the operation of the network or violates the rules and
335 | protocols for communication across the network.
336 |
337 | Corresponding Source conveyed, and Installation Information provided,
338 | in accord with this section must be in a format that is publicly
339 | documented (and with an implementation available to the public in
340 | source code form), and must require no special password or key for
341 | unpacking, reading or copying.
342 |
343 | 7. Additional Terms.
344 |
345 | "Additional permissions" are terms that supplement the terms of this
346 | License by making exceptions from one or more of its conditions.
347 | Additional permissions that are applicable to the entire Program shall
348 | be treated as though they were included in this License, to the extent
349 | that they are valid under applicable law. If additional permissions
350 | apply only to part of the Program, that part may be used separately
351 | under those permissions, but the entire Program remains governed by
352 | this License without regard to the additional permissions.
353 |
354 | When you convey a copy of a covered work, you may at your option
355 | remove any additional permissions from that copy, or from any part of
356 | it. (Additional permissions may be written to require their own
357 | removal in certain cases when you modify the work.) You may place
358 | additional permissions on material, added by you to a covered work,
359 | for which you have or can give appropriate copyright permission.
360 |
361 | Notwithstanding any other provision of this License, for material you
362 | add to a covered work, you may (if authorized by the copyright holders of
363 | that material) supplement the terms of this License with terms:
364 |
365 | a) Disclaiming warranty or limiting liability differently from the
366 | terms of sections 15 and 16 of this License; or
367 |
368 | b) Requiring preservation of specified reasonable legal notices or
369 | author attributions in that material or in the Appropriate Legal
370 | Notices displayed by works containing it; or
371 |
372 | c) Prohibiting misrepresentation of the origin of that material, or
373 | requiring that modified versions of such material be marked in
374 | reasonable ways as different from the original version; or
375 |
376 | d) Limiting the use for publicity purposes of names of licensors or
377 | authors of the material; or
378 |
379 | e) Declining to grant rights under trademark law for use of some
380 | trade names, trademarks, or service marks; or
381 |
382 | f) Requiring indemnification of licensors and authors of that
383 | material by anyone who conveys the material (or modified versions of
384 | it) with contractual assumptions of liability to the recipient, for
385 | any liability that these contractual assumptions directly impose on
386 | those licensors and authors.
387 |
388 | All other non-permissive additional terms are considered "further
389 | restrictions" within the meaning of section 10. If the Program as you
390 | received it, or any part of it, contains a notice stating that it is
391 | governed by this License along with a term that is a further
392 | restriction, you may remove that term. If a license document contains
393 | a further restriction but permits relicensing or conveying under this
394 | License, you may add to a covered work material governed by the terms
395 | of that license document, provided that the further restriction does
396 | not survive such relicensing or conveying.
397 |
398 | If you add terms to a covered work in accord with this section, you
399 | must place, in the relevant source files, a statement of the
400 | additional terms that apply to those files, or a notice indicating
401 | where to find the applicable terms.
402 |
403 | Additional terms, permissive or non-permissive, may be stated in the
404 | form of a separately written license, or stated as exceptions;
405 | the above requirements apply either way.
406 |
407 | 8. Termination.
408 |
409 | You may not propagate or modify a covered work except as expressly
410 | provided under this License. Any attempt otherwise to propagate or
411 | modify it is void, and will automatically terminate your rights under
412 | this License (including any patent licenses granted under the third
413 | paragraph of section 11).
414 |
415 | However, if you cease all violation of this License, then your
416 | license from a particular copyright holder is reinstated (a)
417 | provisionally, unless and until the copyright holder explicitly and
418 | finally terminates your license, and (b) permanently, if the copyright
419 | holder fails to notify you of the violation by some reasonable means
420 | prior to 60 days after the cessation.
421 |
422 | Moreover, your license from a particular copyright holder is
423 | reinstated permanently if the copyright holder notifies you of the
424 | violation by some reasonable means, this is the first time you have
425 | received notice of violation of this License (for any work) from that
426 | copyright holder, and you cure the violation prior to 30 days after
427 | your receipt of the notice.
428 |
429 | Termination of your rights under this section does not terminate the
430 | licenses of parties who have received copies or rights from you under
431 | this License. If your rights have been terminated and not permanently
432 | reinstated, you do not qualify to receive new licenses for the same
433 | material under section 10.
434 |
435 | 9. Acceptance Not Required for Having Copies.
436 |
437 | You are not required to accept this License in order to receive or
438 | run a copy of the Program. Ancillary propagation of a covered work
439 | occurring solely as a consequence of using peer-to-peer transmission
440 | to receive a copy likewise does not require acceptance. However,
441 | nothing other than this License grants you permission to propagate or
442 | modify any covered work. These actions infringe copyright if you do
443 | not accept this License. Therefore, by modifying or propagating a
444 | covered work, you indicate your acceptance of this License to do so.
445 |
446 | 10. Automatic Licensing of Downstream Recipients.
447 |
448 | Each time you convey a covered work, the recipient automatically
449 | receives a license from the original licensors, to run, modify and
450 | propagate that work, subject to this License. You are not responsible
451 | for enforcing compliance by third parties with this License.
452 |
453 | An "entity transaction" is a transaction transferring control of an
454 | organization, or substantially all assets of one, or subdividing an
455 | organization, or merging organizations. If propagation of a covered
456 | work results from an entity transaction, each party to that
457 | transaction who receives a copy of the work also receives whatever
458 | licenses to the work the party's predecessor in interest had or could
459 | give under the previous paragraph, plus a right to possession of the
460 | Corresponding Source of the work from the predecessor in interest, if
461 | the predecessor has it or can get it with reasonable efforts.
462 |
463 | You may not impose any further restrictions on the exercise of the
464 | rights granted or affirmed under this License. For example, you may
465 | not impose a license fee, royalty, or other charge for exercise of
466 | rights granted under this License, and you may not initiate litigation
467 | (including a cross-claim or counterclaim in a lawsuit) alleging that
468 | any patent claim is infringed by making, using, selling, offering for
469 | sale, or importing the Program or any portion of it.
470 |
471 | 11. Patents.
472 |
473 | A "contributor" is a copyright holder who authorizes use under this
474 | License of the Program or a work on which the Program is based. The
475 | work thus licensed is called the contributor's "contributor version".
476 |
477 | A contributor's "essential patent claims" are all patent claims
478 | owned or controlled by the contributor, whether already acquired or
479 | hereafter acquired, that would be infringed by some manner, permitted
480 | by this License, of making, using, or selling its contributor version,
481 | but do not include claims that would be infringed only as a
482 | consequence of further modification of the contributor version. For
483 | purposes of this definition, "control" includes the right to grant
484 | patent sublicenses in a manner consistent with the requirements of
485 | this License.
486 |
487 | Each contributor grants you a non-exclusive, worldwide, royalty-free
488 | patent license under the contributor's essential patent claims, to
489 | make, use, sell, offer for sale, import and otherwise run, modify and
490 | propagate the contents of its contributor version.
491 |
492 | In the following three paragraphs, a "patent license" is any express
493 | agreement or commitment, however denominated, not to enforce a patent
494 | (such as an express permission to practice a patent or covenant not to
495 | sue for patent infringement). To "grant" such a patent license to a
496 | party means to make such an agreement or commitment not to enforce a
497 | patent against the party.
498 |
499 | If you convey a covered work, knowingly relying on a patent license,
500 | and the Corresponding Source of the work is not available for anyone
501 | to copy, free of charge and under the terms of this License, through a
502 | publicly available network server or other readily accessible means,
503 | then you must either (1) cause the Corresponding Source to be so
504 | available, or (2) arrange to deprive yourself of the benefit of the
505 | patent license for this particular work, or (3) arrange, in a manner
506 | consistent with the requirements of this License, to extend the patent
507 | license to downstream recipients. "Knowingly relying" means you have
508 | actual knowledge that, but for the patent license, your conveying the
509 | covered work in a country, or your recipient's use of the covered work
510 | in a country, would infringe one or more identifiable patents in that
511 | country that you have reason to believe are valid.
512 |
513 | If, pursuant to or in connection with a single transaction or
514 | arrangement, you convey, or propagate by procuring conveyance of, a
515 | covered work, and grant a patent license to some of the parties
516 | receiving the covered work authorizing them to use, propagate, modify
517 | or convey a specific copy of the covered work, then the patent license
518 | you grant is automatically extended to all recipients of the covered
519 | work and works based on it.
520 |
521 | A patent license is "discriminatory" if it does not include within
522 | the scope of its coverage, prohibits the exercise of, or is
523 | conditioned on the non-exercise of one or more of the rights that are
524 | specifically granted under this License. You may not convey a covered
525 | work if you are a party to an arrangement with a third party that is
526 | in the business of distributing software, under which you make payment
527 | to the third party based on the extent of your activity of conveying
528 | the work, and under which the third party grants, to any of the
529 | parties who would receive the covered work from you, a discriminatory
530 | patent license (a) in connection with copies of the covered work
531 | conveyed by you (or copies made from those copies), or (b) primarily
532 | for and in connection with specific products or compilations that
533 | contain the covered work, unless you entered into that arrangement,
534 | or that patent license was granted, prior to 28 March 2007.
535 |
536 | Nothing in this License shall be construed as excluding or limiting
537 | any implied license or other defenses to infringement that may
538 | otherwise be available to you under applicable patent law.
539 |
540 | 12. No Surrender of Others' Freedom.
541 |
542 | If conditions are imposed on you (whether by court order, agreement or
543 | otherwise) that contradict the conditions of this License, they do not
544 | excuse you from the conditions of this License. If you cannot convey a
545 | covered work so as to satisfy simultaneously your obligations under this
546 | License and any other pertinent obligations, then as a consequence you may
547 | not convey it at all. For example, if you agree to terms that obligate you
548 | to collect a royalty for further conveying from those to whom you convey
549 | the Program, the only way you could satisfy both those terms and this
550 | License would be to refrain entirely from conveying the Program.
551 |
552 | 13. Use with the GNU Affero General Public License.
553 |
554 | Notwithstanding any other provision of this License, you have
555 | permission to link or combine any covered work with a work licensed
556 | under version 3 of the GNU Affero General Public License into a single
557 | combined work, and to convey the resulting work. The terms of this
558 | License will continue to apply to the part which is the covered work,
559 | but the special requirements of the GNU Affero General Public License,
560 | section 13, concerning interaction through a network will apply to the
561 | combination as such.
562 |
563 | 14. Revised Versions of this License.
564 |
565 | The Free Software Foundation may publish revised and/or new versions of
566 | the GNU General Public License from time to time. Such new versions will
567 | be similar in spirit to the present version, but may differ in detail to
568 | address new problems or concerns.
569 |
570 | Each version is given a distinguishing version number. If the
571 | Program specifies that a certain numbered version of the GNU General
572 | Public License "or any later version" applies to it, you have the
573 | option of following the terms and conditions either of that numbered
574 | version or of any later version published by the Free Software
575 | Foundation. If the Program does not specify a version number of the
576 | GNU General Public License, you may choose any version ever published
577 | by the Free Software Foundation.
578 |
579 | If the Program specifies that a proxy can decide which future
580 | versions of the GNU General Public License can be used, that proxy's
581 | public statement of acceptance of a version permanently authorizes you
582 | to choose that version for the Program.
583 |
584 | Later license versions may give you additional or different
585 | permissions. However, no additional obligations are imposed on any
586 | author or copyright holder as a result of your choosing to follow a
587 | later version.
588 |
589 | 15. Disclaimer of Warranty.
590 |
591 | THERE IS NO WARRANTY FOR THE PROGRAM, TO THE EXTENT PERMITTED BY
592 | APPLICABLE LAW. EXCEPT WHEN OTHERWISE STATED IN WRITING THE COPYRIGHT
593 | HOLDERS AND/OR OTHER PARTIES PROVIDE THE PROGRAM "AS IS" WITHOUT WARRANTY
594 | OF ANY KIND, EITHER EXPRESSED OR IMPLIED, INCLUDING, BUT NOT LIMITED TO,
595 | THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR
596 | PURPOSE. THE ENTIRE RISK AS TO THE QUALITY AND PERFORMANCE OF THE PROGRAM
597 | IS WITH YOU. SHOULD THE PROGRAM PROVE DEFECTIVE, YOU ASSUME THE COST OF
598 | ALL NECESSARY SERVICING, REPAIR OR CORRECTION.
599 |
600 | 16. Limitation of Liability.
601 |
602 | IN NO EVENT UNLESS REQUIRED BY APPLICABLE LAW OR AGREED TO IN WRITING
603 | WILL ANY COPYRIGHT HOLDER, OR ANY OTHER PARTY WHO MODIFIES AND/OR CONVEYS
604 | THE PROGRAM AS PERMITTED ABOVE, BE LIABLE TO YOU FOR DAMAGES, INCLUDING ANY
605 | GENERAL, SPECIAL, INCIDENTAL OR CONSEQUENTIAL DAMAGES ARISING OUT OF THE
606 | USE OR INABILITY TO USE THE PROGRAM (INCLUDING BUT NOT LIMITED TO LOSS OF
607 | DATA OR DATA BEING RENDERED INACCURATE OR LOSSES SUSTAINED BY YOU OR THIRD
608 | PARTIES OR A FAILURE OF THE PROGRAM TO OPERATE WITH ANY OTHER PROGRAMS),
609 | EVEN IF SUCH HOLDER OR OTHER PARTY HAS BEEN ADVISED OF THE POSSIBILITY OF
610 | SUCH DAMAGES.
611 |
612 | 17. Interpretation of Sections 15 and 16.
613 |
614 | If the disclaimer of warranty and limitation of liability provided
615 | above cannot be given local legal effect according to their terms,
616 | reviewing courts shall apply local law that most closely approximates
617 | an absolute waiver of all civil liability in connection with the
618 | Program, unless a warranty or assumption of liability accompanies a
619 | copy of the Program in return for a fee.
620 |
621 | END OF TERMS AND CONDITIONS
622 |
623 | How to Apply These Terms to Your New Programs
624 |
625 | If you develop a new program, and you want it to be of the greatest
626 | possible use to the public, the best way to achieve this is to make it
627 | free software which everyone can redistribute and change under these terms.
628 |
629 | To do so, attach the following notices to the program. It is safest
630 | to attach them to the start of each source file to most effectively
631 | state the exclusion of warranty; and each file should have at least
632 | the "copyright" line and a pointer to where the full notice is found.
633 |
634 |
635 | Copyright (C)
636 |
637 | This program is free software: you can redistribute it and/or modify
638 | it under the terms of the GNU General Public License as published by
639 | the Free Software Foundation, either version 3 of the License, or
640 | (at your option) any later version.
641 |
642 | This program is distributed in the hope that it will be useful,
643 | but WITHOUT ANY WARRANTY; without even the implied warranty of
644 | MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
645 | GNU General Public License for more details.
646 |
647 | You should have received a copy of the GNU General Public License
648 | along with this program. If not, see .
649 |
650 | Also add information on how to contact you by electronic and paper mail.
651 |
652 | If the program does terminal interaction, make it output a short
653 | notice like this when it starts in an interactive mode:
654 |
655 | Copyright (C)
656 | This program comes with ABSOLUTELY NO WARRANTY; for details type `show w'.
657 | This is free software, and you are welcome to redistribute it
658 | under certain conditions; type `show c' for details.
659 |
660 | The hypothetical commands `show w' and `show c' should show the appropriate
661 | parts of the General Public License. Of course, your program's commands
662 | might be different; for a GUI interface, you would use an "about box".
663 |
664 | You should also get your employer (if you work as a programmer) or school,
665 | if any, to sign a "copyright disclaimer" for the program, if necessary.
666 | For more information on this, and how to apply and follow the GNU GPL, see
667 | .
668 |
669 | The GNU General Public License does not permit incorporating your program
670 | into proprietary programs. If your program is a subroutine library, you
671 | may consider it more useful to permit linking proprietary applications with
672 | the library. If this is what you want to do, use the GNU Lesser General
673 | Public License instead of this License. But first, please read
674 | .
675 |
--------------------------------------------------------------------------------
/README.md:
--------------------------------------------------------------------------------
1 | # Single-cell eQTL analysis of memory T cells
2 |
3 | ## Overview
4 | This repo provides code used for single-cell eQTL analysis of single-cell CITE-seq data from 500,089 memory T cells. If you are interested in code for any additional analyses from our study, please contact us at aparnanathan@g.harvard.edu
5 |
6 | This study is available as a preprint:
7 |
8 | > Nathan A, et al, 2021. Modeling memory T cell states at single-cell resolution identifies in vivo state-dependence of eQTLs influencing disease. [bioRxiv](https://doi.org/10.1101/2021.07.29.454316).
9 |
10 | ## Data availability
11 |
12 | Single-cell data are availale on GEO (Accession: GSE158769) and can be interactively browsed on [our website](https://immunogenomics.io/tbru/).
13 | Genotyping data are available on dbGaP (Accession: phs002025).
14 |
15 | Please send any questions regarding the study, data, or analyses to Aparna Nathan (aparnanathan@g.harvard.edu).
16 |
--------------------------------------------------------------------------------
/scripts/external_data/Randolph2021/mock_pbmc_processing.R:
--------------------------------------------------------------------------------
1 | # Scripts to process mock-treated PBMC single-cell data from Randolph, et al. (2021) Science
2 |
3 | # Load libraries
4 |
5 | library(Seurat)
6 | library(Matrix)
7 | library(singlecellmethods)
8 | library(dplyr)
9 | library(harmony)
10 | library(uwot)
11 |
12 | # Load data
13 | ge_counts <- readRDS('/path/randolph_raw.rds') # 19248 236993 (genes x cells)
14 | meta.data <- readRDS('/path/randolph_meta.rds') # 236993 17
15 |
16 | # Remove IAV genes from Gene count matrix
17 | iav_genes <- c('PB2','PB1','PA','HA','NP','NA','M2','M1','NEP','NS1')
18 | ge_counts <- ge_counts[!rownames(ge_counts) %in% iav_genes,]
19 |
20 | # Select MOCK
21 | meta.data <- meta.data[meta.data$SOC_infection_status=='NI',] # 124976 cells
22 | ge_counts <- ge_counts[,colnames(ge_counts) %in% rownames(meta.data)]
23 |
24 | # Select cells with > 500 genes
25 | meta.data <- meta.data[meta.data$nFeature_RNA>500,] # 108,209 18
26 | ge_counts <- ge_counts[ ,colnames(ge_counts) %in% rownames(meta.data)]
27 |
28 | # Normalize and scale variable genes
29 | mrnaNorm <- singlecellmethods::normalizeData(ge_counts, method = "log")
30 | var_genes <- singlecellmethods::vargenes_vst(object = ge_counts, topn = 3000)
31 | mrnaCosScaled <- mrnaNorm[rownames(mrnaNorm) %in% var_genes, ] %>% scaleData() %>% cosine_normalize(2)
32 |
33 | # PCA
34 | pca_mrnaScaled <- irlba::prcomp_irlba(t(mrnaCosScaled ), 20)
35 |
36 | # Harmony batch correction
37 | harmony <- HarmonyMatrix(pca_mrnaScaled$x, meta.data, c("SOC_indiv_ID","batchID"), theta = c(1,1), lambda = c(1,1),
38 | plot_convergence = TRUE, nclust = 100, max.iter.harmony = 20,
39 | max.iter.cluster = 20, do_pca = F, verbose = T)
40 |
41 | # UMAP
42 | umap_mrnaScaled <- uwot::umap(harmony, n_neighbors = 30, metric = "euclidean", min_dist = .1)
43 |
--------------------------------------------------------------------------------
/scripts/external_data/Randolph2021/monocytes_trajectory.R:
--------------------------------------------------------------------------------
1 | # Scripts to conduct trajectory analysis on monocytes from Randolph, et al. (2021) Science
2 |
3 | # Load libraries
4 |
5 | library(Seurat)
6 | library(Matrix)
7 | library(singlecellmethods)
8 | library(dplyr)
9 | library(harmony)
10 | library(uwot)
11 | library(DDRTree)
12 | library(princurve)
13 |
14 | # Load post-QC data
15 |
16 | exprs_raw <- readRDS("/path/randolph_raw.rds")
17 | exprs_raw <- exprs_raw[!row.names(exprs_raw) %in% c("PB2", "PB1", "PA", "HA", "NP", "NA", "M2", "M1", "NEP", "NS1"),]
18 | meta.data <- readRDS("/path/randolph_meta.rds")
19 |
20 | # Normalize and scale
21 |
22 | exprs_raw <- exprs_raw[,row.names(meta.data)][,grepl("monocyte", meta.data$celltype)]
23 | mono <- CreateSeuratObject(counts = exprs_raw)
24 | mono <- PercentageFeatureSet(mono, pattern = "^MT-", col.name = "percent.mt")
25 | mono <- SCTransform(mono, vars.to.regress = "percent.mt", verbose = FALSE)
26 |
27 | # PCA
28 |
29 | pca_res_filter <- irlba::prcomp_irlba(t(mono@assays$SCT@scale.data), 20)
30 |
31 | # Batch correction with Harmony
32 |
33 | harmony_res <- HarmonyMatrix(pca_res_filter$x, meta.data[grepl("monocyte", meta.data$celltype),], c("SOC_indiv_ID", "batchID", "SOC_infection_status"), theta = c(1,1,1),
34 | plot_convergence = TRUE, max.iter.harmony = 10, epsilon.cluster = -Inf, epsilon.harmony = -Inf,
35 | max.iter.cluster = 10, do_pca = F, verbose = T, return_object = FALSE)
36 |
37 | # UMAP for visualization
38 |
39 | umap_res_harmony <- umap(harmony_res[,1:10], n_neighbors = 30L, metric = "euclidean", min_dist = .1)
40 |
41 | # Plot cells by treatment condition
42 |
43 | fig.size(4,5)
44 | ggplot(as.data.frame(umap_res_harmony), aes(V1, V2, color = meta.data[grepl("monocyte", meta.data$celltype), "SOC_infection_status"])) +
45 | geom_point(shape = ".") + theme_classic() + theme(legend.title = element_blank(), axis.text = element_blank(), axis.title = element_text(size = 12)) +
46 | xlab("UMAP1") + ylab("UMAP2")
47 |
48 | # Trajectory analysis with DDRTree
49 |
50 | ncells <- nrow(harmony_res)
51 | ncenter <- round(2 * 100 * log(ncells) / (log(ncells) + log(100)))
52 |
53 | ddr_args <- c(list(
54 | X = t(harmony_res),
55 | dimensions = 2, ## LOW DIMENSIONALITY
56 | ncenter = ncenter, ## number of nodes allowed in the regularization graph
57 | param.gamma = 10, ## param.gamma regularization parameter for k-means
58 | maxIter = 20,
59 | tol = 1e-3,
60 | sigma = 0.0001,
61 | verbose = FALSE))
62 |
63 | ddrtree_res <<- do.call(DDRTree, ddr_args)
64 | pseud_integr <- t(ddrtree_res$Z)
65 | pseud_integr <- as.data.frame(pseud_integr)
66 |
67 | pc_res <- principal_curve(t(ddrtree_res$Z))
68 | pseud_integr$pseudo <- pc_res$lambda
69 |
70 | # Plot trajectory across cells
71 |
72 | fig.size(4,5)
73 | ggplot(data = as.data.frame(umap_res_harmony), aes(x = V1, y = V2, color = pseud_integr$pseudo)) +
74 | geom_point(shape = ".") +
75 | theme_classic() +
76 | theme(legend.text=element_text(size=12)) + scale_color_viridis(option = "plasma") +
77 | theme(legend.title=element_blank(), axis.text = element_blank()) +
78 | xlab("UMAP1") + ylab("UMAP2")
79 |
80 | # Plot deciles by treatment condition composition
81 | data.frame(V3 = meta.data.mono$SOC_infection_status) %>% mutate(bin = cut(pseud_integr$pseudo, breaks = quantile(pseud_integr$pseudo, probs = seq(0, 1, .1), labels = 1:10, include.lowest = TRUE))) %>%
82 | # group_by(bin, V3) %>% summarise(count = n()) %>% filter(!is.na(bin)) %>%
83 | filter(!is.na(bin)) %>% ggplot(aes(bin, fill = V3)) + geom_bar(stat = "count", position = "fill") + theme_classic() +
84 | theme(axis.text = element_text(size = 12), axis.title = element_text(size = 14))
85 |
--------------------------------------------------------------------------------
/scripts/external_data/Smillie2019/singlecell_analysis.R:
--------------------------------------------------------------------------------
1 | # Script to process single-cell data from Smillie, et al. (2019) Cell
2 |
3 | # Load libraries
4 |
5 | library(Seurat)
6 | library(Matrix)
7 | library(singlecellmethods)
8 | library(dplyr)
9 | library(harmony)
10 | library(uwot)
11 |
12 | # Load data
13 |
14 | data <- readMM("/path/gene_sorted-Imm.matrix.mtx")
15 | genes <- read.table("/path/Imm.genes.tsv")
16 | row.names(data) <- genes$V1
17 |
18 | cells <- read.table("/path/Imm.barcodes2.tsv")
19 | colnames(data) <- cells$V1
20 |
21 | meta_data <- read.table("/path/all.meta2.txt", sep = "\t", header = T)
22 | meta_data <- meta_data[-1,]
23 | row.names(meta_data) <- meta_data$NAME
24 | meta_data$Subject <- factor(meta_data$Subject, levels = levels(meta_data$Subject)[-1])
25 |
26 | # Select T cells
27 |
28 | meta_data$Tcell = meta_data$Cluster %in% c("CD4+ Activated Fos-hi", "CD4+ Activated Fos-lo", "CD4+ Memory", "CD4+ PD1+", "CD8+ IELs",
29 | "CD8+ IL17+", "CD8+ LP", "Cycling T", "Tregs")
30 | meta_data <- meta_data[meta_data$Tcell,]
31 | data <- data[,row.names(meta_data)]
32 |
33 | # QC cells (> 300 genes, < 20% MT UMIs)
34 |
35 | meta_data$percent_mito <- colSums(data[grepl("^MT-", row.names(data)),])/colSums(data)
36 | meta_data$nGene <- as.numeric(as.character(meta_data$nGene))
37 | meta_data <- meta_data[meta_data$nGene > 300 & meta_data$percent_mito < .2,]
38 | data <- data[,row.names(meta_data)]
39 |
40 | # Normalize and scale
41 |
42 | data <- as(data, "dgCMatrix")
43 | var_genes <- vargenes_vst(data, meta_data$Subject, 200)
44 | exprs_norm <- normalizeData(data, method = "log")
45 | cc_genes <- c(Seurat::cc.genes$s.genes, Seurat::cc.genes$g2m.genes)
46 | exprs_cosine <- exprs_norm[var_genes[!var_genes %in% cc_genes],] %>% scaleData %>% cosine_normalize(2)
47 |
48 | # PCA
49 |
50 | pca_res_filter <- irlba::prcomp_irlba(t(exprs_cosine), 20)
51 |
52 | # Batch correction with Harmony
53 |
54 | harmony_res <- HarmonyMatrix(pca_res_filter$x, meta_data, c("Subject"), theta = c(1),
55 | plot_convergence = TRUE, max.iter.harmony = 10, epsilon.cluster = -Inf, epsilon.harmony = -Inf,
56 | max.iter.cluster = 10, do_pca = F, verbose = T, return_object = FALSE)
57 |
58 | # UMAP for visualization
59 |
60 | umap_res_harmony <- umap(harmony_res, n_neighbors = 30L, metric = "euclidean", min_dist = .1)
61 |
--------------------------------------------------------------------------------
/scripts/finemap/01_calc_LD.sh:
--------------------------------------------------------------------------------
1 | # Calculate in-sample LD between variants
2 |
3 | chr=$1
4 | gene=$2
5 |
6 |
7 | /path/plinkv1.90b6.22 --vcf merged_${chr}.QC.vcf.gz \
8 | --extract merged_${chr}_${gene}.TMP.snpIDs \
9 | --keep-allele-order \
10 | --r square \
11 | --out merged_${chr}_${gene}.TMP
12 |
13 | /path/plinkv1.90b6.22 --vcf merged_${chr}.QC.vcf.gz \
14 | --extract merged_${chr}_${gene}.TMP.snpIDs \
15 | --keep-allele-order \
16 | --freq \
17 | --out merged_${chr}_${gene}.TMP
18 |
19 | sed -i 's/\t/ /g' merged_${chr}_${gene}.TMP.ld
20 | sed -i 's/nan/0/g' merged_${chr}_${gene}.TMP.ld
21 |
--------------------------------------------------------------------------------
/scripts/finemap/01_normalize_blueprint.R:
--------------------------------------------------------------------------------
1 | # Script to process Blueprint data similarly to pseudobulk memory T cell data by PEER normalizing and regressing out covariates
2 |
3 | library(data.table)
4 | library(readr)
5 | library(stringr)
6 | library(peer)
7 |
8 | K = 30
9 |
10 | ### phenotype data downloaded from here: http://dcc.blueprint-epigenome.eu/#/datasets/EGAD00001002671
11 | pheno <- read.delim("/path/EGAD00001002663.pheno.txt", sep = "\t")
12 | pheno <- pheno[c("DONOR_ID","DONOR_AGE","DONOR_SEX")]
13 | pheno <- pheno[!duplicated(pheno$DONOR_ID),]
14 | row.names(pheno) <- pheno$DONOR_ID
15 | pheno <- pheno[c("DONOR_AGE","DONOR_SEX")]
16 | pheno <- as.data.frame(pheno)
17 |
18 | ### Expression data (batch corrected and normalized)
19 | norm_exp <- fread(paste0("/path/tcel_gene_nor_combat_20151109.txt.gz"))
20 | norm_exp <- as.data.frame(norm_exp)
21 | rownames(norm_exp) <- str_split_fixed(norm_exp$ens.id, "[.]", 2)[,1]
22 | norm_exp <- norm_exp[,-1]
23 | dim(norm_exp)
24 |
25 | ### select the genes that are also present in single-cell data
26 | tbru_exp <- read_tsv("/path/tbru_IDs.txt.gz")
27 | tbru_exp <- as.data.frame(tbru_exp)
28 | norm_exp <- subset(norm_exp, rownames(norm_exp) %in% tbru_exp$ids)
29 |
30 | ### inverse normal transformation
31 | rn<-apply(norm_exp,1,function(x){
32 | qnorm( (rank(x, na.last="keep") - 0.5) / sum(!is.na(x)) )
33 | })
34 |
35 | ### Genotyping PCs
36 | pcs <- read.delim("/path/EGAZ00001235598_blueprint06092016_qc_pruned_noMHC.pc.eigenvec", sep = " ", header = F)
37 | row.names(pcs) <- pcs$V2
38 | pcs <- pcs[,3:ncol(pcs)]
39 | colnames(pcs) <- c(sprintf("gPC%d", seq(1,20)))
40 | pcs <- pcs[c("gPC1", "gPC2", "gPC3")]
41 |
42 | ### age group and sex as covariates
43 | covs <- merge(pheno, pcs, by='row.names')
44 | rownames(covs) <- covs[,1]
45 | covs <- covs[,-1]
46 | covs <- covs[rownames(covs) %in% rownames(rn),]
47 | covs <- covs[match(rownames(rn), rownames(covs)),]
48 | levels(covs$DONOR_AGE) <- 1:10
49 | covs$DONOR_SEX <- covs$DONOR_SEX == "Female"
50 | covs$DONOR_AGE <- as.numeric(covs$DONOR_AGE)
51 | head(covs)
52 |
53 | ### correct for covariates and 30 peer factors
54 | model = PEER()
55 | PEER_setPhenoMean(model, as.matrix(rn))
56 | dim(PEER_getPhenoMean(model))
57 | PEER_setAdd_mean(model, TRUE)
58 | PEER_setNk(model,K)
59 | PEER_getNk(model)
60 | PEER_setCovariates(model, as.matrix(covs))
61 | PEER_setNmax_iterations(model,10000)
62 |
63 | #perform the inference
64 | PEER_update(model)
65 |
66 | ### Output the peer factors
67 | factors = PEER_getX(model)
68 | dump <- data.frame(id=colnames(norm_exp),factors)
69 | gz1 <- gzfile(paste0("/path/Blueprint_INT_NoePCs_3gPCs_age_sex_peer_factors_K30.txt.gz"),"w")
70 | write.table(dump, gz1, sep = "\t", quote = F, row.names = FALSE)
71 | close(gz1)
72 |
73 | ### output the residulas
74 | residuals = PEER_getResiduals(model)
75 | colnames(residuals) <- row.names(norm_exp)
76 | row.names(residuals) <- colnames(norm_exp)
77 | dump <- data.frame( ids = row.names(norm_exp),t(residuals))
78 | gz1 <- gzfile(paste0("/path/Blueprint_INT_NoePCs_3gPCs_age_sex_peer_factors_K30_residuals.txt.gz"),"w")
79 | write.table(dump, gz1, sep = "\t", quote = F, row.names = FALSE)
80 | close(gz1)
81 |
--------------------------------------------------------------------------------
/scripts/finemap/02_merge_pseudobulk_blueprint_snps.sh:
--------------------------------------------------------------------------------
1 | #!/bin/bash
2 |
3 | # Script to merge genotypes from pseudobulk memory T cells and Blueprint
4 |
5 | module load vcftools/0.1.15
6 | module load tabix/0.2.6
7 | module load bcftools/1.10.2
8 |
9 | # Format pseudobulk genotype VCF
10 | bcftools annotate --set-id '%CHROM\_%POS\_%REF\_%FIRST_ALT' /path/geno_imputed_dosage_maf05_hg38_${chr}.vcf.gz | bgzip -c > /path/geno_imputed_dosage_maf05_hg38_${chr}_QC.vcf.gz
11 | tabix -f -p /path/geno_imputed_dosage_maf05_hg38_${chr}_QC.vcf.gz
12 |
13 | # Format Blueprint genotype VCF
14 | bcftools annotate --set-id '%CHROM\_%POS\_%REF\_%FIRST_ALT' /path/EGAZ00001235598_blueprint06092016_qc_hg38_PICARD_chr${chr}.vcf.gz | bgzip -c > /path/EGAZ00001235598_blueprint06092016_qc_hg38_PICARD_chr${chr}.QC.vcf.gz
15 | tabix -f -p vcf BLUEPRINT/PlinkFiles/EGAZ00001235598_blueprint06092016_qc_hg38_PICARD_chr${chr}.QC.vcf.gz
16 |
17 | # Merge VCFs
18 | /path/vcftools/0.1.15/bin/vcf-merge /path/EGAZ00001235598_blueprint06092016_qc_hg38_PICARD_chr${chr}.QC.vcf.gz /path/geno_imputed_dosage_maf05_hg38_${chr}_QC.vcf.gz | bgzip -c > /path/merged_chr${chr}.vcf.gz
19 | tabix -f -p vcf /path/merged_chr${chr}.vcf.gz
20 |
21 | # Get pseudobulk variants
22 | zcat /path/geno_imputed_dosage_maf05_hg38_${chr}_QC.vcf.gz | grep -v "#" | cut -f3,3 | sort | uniq -c | awk '{if (\$1 == 1) print \$2}' > /path/pseudobulk_chr${chr}.VarIDs.txt
23 |
24 | # Blueprint variants
25 | zcat /path/EGAZ00001235598_blueprint06092016_qc_hg38_PICARD_chr${chr}.QC.vcf.gz | grep -v "#" | cut -f3,3 > /path/Blueprint_chr${chr}.VarIDs.txt
26 |
27 | # Get shared variant IDs
28 | cat /path/pseudobulk_chr${chr}.VarIDs.txt /path/Blueprint_chr${chr}.VarIDs.txt | sort | uniq -c | awk '{if (\$1 == 2) print \$2}' > /merged_chr${chr}.VarIDs.txt
29 |
30 | # Merged VCF of shared variants
31 | /path/vcftools/0.1.15/bin/vcftools --gzvcf /path/merged_chr${chr}.vcf.gz --snps /path/merged_chr${chr}.VarIDs.txt --recode --out /path/merged_chr${chr}.QC --keep /path/merged_QC.vcftools.keep
32 | bgzip -f /path/merged_chr${chr}.QC.vcf
33 | tabix -f -p vcf /path/merged_chr${chr}.QC.vcf.gz
34 |
--------------------------------------------------------------------------------
/scripts/finemap/02_prep_finemap_files.R:
--------------------------------------------------------------------------------
1 | # Script to prepare files for fine-mapping from FastQTL nominal output
2 |
3 | gene = "CTLA4"
4 |
5 | # Read fastqtl nominal output
6 | data <- read_delim(path1, delim=" ", col_names=F)
7 | colnames(data) <- c("EnsemblID", "var", "dist", "P", "beta")
8 | data <- data %>% filter(EnsemblID == gene)
9 |
10 | # Read minor allele frequencies per variant
11 | maf <- read.table(path2, header=T)
12 |
13 | data <- merge(data, maf, by.x="var", by.y="SNP")
14 |
15 | data$Z <- qnorm(data$P/2, lower.tail = FALSE)
16 | data$Z <- ifelse(data$beta < 0, -1*data$Z, data$Z)
17 | data$SE <- data$beta/data$Z
18 |
19 | data <- data[,c("var", "Z")]
20 | data <- unique(data)
21 |
22 | write.table(data, paste(path3,"_", gene, ".TMP.z", sep=""), quote=F, col.names=F, row.names=F)
23 |
--------------------------------------------------------------------------------
/scripts/finemap/03_merge_pseudobulk_blueprint_resid.R:
--------------------------------------------------------------------------------
1 | # Script to merge PEER residuals from pseudobulk memory T cells and Blueprint bulk T cells
2 |
3 | library(data.table)
4 | library(stringr)
5 | library(readr)
6 |
7 | ## get GTF files of TSS for each gene in Cell Ranger reference (refdata-cellranger-GRCh38-3.0.0)
8 | files <- Sys.glob(file.path("/path/chr*_tss_gene_basic.bed"))
9 |
10 | tss <- rbindlist(lapply(files, function(filename) {
11 | read.table(filename, header=T)
12 | }), fill=TRUE)
13 |
14 | tss$CHR <- str_remove(tss$CHR, "chr")
15 | tss$ENS <- str_split_fixed(tss$ID, "[.]", 2)[,1]
16 | tss$start <- tss$POS-1
17 |
18 | tss <- tss[,c(1,6,4,5)]
19 | colnames(tss) <- c("CHR", "Start", "End", "ID")
20 |
21 | ## combine residuals from the two datasets -> 13978 genes
22 | tbru_exp <- read_tsv("/path/peer_residuals.txt.gz")
23 | tbru_exp <- as.data.frame(tbru_exp)
24 | rownames(tbru_exp) <- tbru_exp$ids
25 | tbru_exp <- tbru_exp[,-1]
26 | blue_exp <- read_tsv("/path/Blueprint_INT_NoePCs_3gPCs_age_sex_peer_factors_K30_residuals.txt.gz")
27 | blue_exp <- as.data.frame(blue_exp)
28 | rownames(blue_exp) <- blue_exp$ids
29 | blue_exp <- blue_exp[,-1]
30 |
31 | merged_exp <- merge(tbru_exp, blue_exp, by='row.names')
32 | merged_exp <- as.data.frame(merged_exp)
33 | dim(merged_exp)
34 | merged_exp[1:4,1:5]
35 |
36 | ### Merge with TSS data from Cell Ranger GTF -> 13963 genes
37 | data <- merge(tss, merged_exp, by.x="ID", by.y="Row.names")
38 | data <- data[with(data, order(as.numeric(CHR), as.numeric(End))),]
39 | data <- as.data.frame(data)
40 | data <- data[,c(2,3,4,1,5:ncol(data))]
41 |
42 | ### fix names to match vcf
43 | colnames(data) <- str_replace_all(colnames(data), "[.]", "-")
44 |
45 | ### create a cohort covariate
46 | #fam <- read.table("/path/SampleIDs.txt", header=F)
47 | #fam <- fam %>% dplyr::mutate(cov=ifelse(grepl('-',V1), 1, 0))
48 |
49 | ## order expression like fam
50 | samp_names <- c(colnames(data[,1:4]), as.character(fam$V1))
51 | data <- data[,match(samp_names, colnames(data))]
52 | data[,1] <- paste("chr", data[,1], sep="")
53 | colnames(data)[1] <- "#CHR"
54 |
55 | reg_exp <- sapply(1:nrow(data), function(x){resid(lm(unlist(data[x,-c(1:4)])~ fam$cov))})
56 | res <- data.frame(data[,1:4],t(reg_exp))
57 | colnames(res)[1] <- "#CHR"
58 | colnames(res) <- str_replace_all(colnames(res), "[.]", "-")
59 | write.table(res,"/path/merged.bed", sep = "\t", quote = F, row.names = FALSE)
60 |
61 | ##bgzip and index
62 |
--------------------------------------------------------------------------------
/scripts/finemap/03_run_finemap.sh:
--------------------------------------------------------------------------------
1 | # Run fine-mapping with CAVIAR
2 |
3 | CAVIAR -o merged_${chr}_${gene}_SingleCausal.caviar -l merged_${chr}_${gene}.TMP.ld -z merged_${chr}_${gene}.TMP.z -c 1
4 |
--------------------------------------------------------------------------------
/scripts/post_analysis/calc_singlecell_betas.R:
--------------------------------------------------------------------------------
1 | # Script to calculate single-cell betas from single-cell eQTL model interaction effects and CV scores
2 |
3 | # Load libraries
4 | library(ggplot2)
5 | library(dplyr)
6 | library(RColorBrewer)
7 |
8 | # Load eQTL results, CV scores, and UMAP coordinates
9 |
10 | sceqtl_results <- readRDS("/path/sceqtl_results.rds")
11 | cv_scores <- readRDS("/path/cca_donor_batch.rds")[,1:7]
12 | umap_res <- readRDS("/path/umap_donor_batch.rds")
13 |
14 | # Calculate single-cell betas
15 |
16 | betas <- sceqtl_results %>% filter(grepl("^G", term) & gene == "GZMA") %>% select(Estimate) %>% unlist
17 | sc_betas <- t(data.frame(X1 = rowSums(sweep(cbind(1, cv_scores[,1:7]), MARGIN=2, betas, `*`))))
18 |
19 | # Plot eQTL betas per cell
20 |
21 | plot_eqtl <- function (ab, umap, exprs, pct = 0.95, geno)
22 | {
23 | max.cutoff = quantile(exprs[ab, ], pct)
24 | min.cutoff = quantile(exprs[ab, ], 1 - pct)
25 | tmp <- sapply(X = exprs[ab, ], FUN = function(x) {
26 | return(ifelse(test = x > max.cutoff, yes = max.cutoff,
27 | no = x))
28 | })
29 | tmp <- sapply(X = tmp, FUN = function(x) {
30 | return(ifelse(test = x < min.cutoff, yes = min.cutoff,
31 | no = x))
32 | })
33 | umap_res_plot <- cbind(umap, tmp)
34 | return(ggplot(data = as.data.frame(umap_res_plot)[sample(nrow(umap_res_plot)),
35 | ], aes(x = V1, y = V2)) + geom_point_rast(mapping = aes(color = tmp),
36 | shape = ".") + scale_color_distiller(palette = "RdYlBu",
37 | limits = c(geno - max(abs(geno - tmp)), geno + max(abs(geno -
38 | tmp)))) + theme_classic() + theme(axis.text = element_blank(),
39 | axis.title = element_blank()))
40 | }
41 |
42 | plot_eqtl("X1", umap_res, sc_betas, 1, betas[1])
43 |
--------------------------------------------------------------------------------
/scripts/post_analysis/cluster_eGenes.R:
--------------------------------------------------------------------------------
1 | # Script to cluster eGenes and test gene set enrichments
2 |
3 | # Load libraries
4 | library(dplyr)
5 | library(tidyr)
6 | library(Seurat)
7 | library(msigdbr)
8 | library(RColorBrewer)
9 | library(pheatmap)
10 | library(parallel)
11 | library(patchwork)
12 |
13 | # Load eQTL output (data frame from concatenating output of single-cell eQTL models; 195,330 rows = 6,511 eGenes x 30 fixed effects estimated for each)
14 | sceqtl_results <- readRDS("/path/sceqtl_results.rds")
15 |
16 | # Extract genotype betas, and interaction betas and p values for 7 eQTL interaction terms per eGene
17 |
18 | g_mat <- sceqtl_results %>% filter(term == "G") %>% select(gene, Estimate)
19 | row.names(g_mat) <- g_mat$gene
20 |
21 | beta_mat <- sceqtl_results %>% filter(grepl("G:CV", term)) %>% select(gene, term, Estimate) %>% spread(term, Estimate)
22 | row.names(beta_mat) <- beta_mat$gene
23 | beta_mat <- beta_mat[,-1]
24 |
25 | p_mat <- sceqtl_results %>% filter(grepl("G:CV", term)) %>% select(gene, term, pval) %>% spread(term, pval)
26 | row.names(p_mat) <- p_mat$gene
27 | p_mat <- p_mat[,-1]
28 |
29 | # Identify significant interactions
30 |
31 | sig_mat <- beta_mat
32 | sig_mat[p_mat >= .05/7/nrow(p_mat)] = 0
33 |
34 | # Adjust beta signs
35 |
36 | beta_mat <- beta_mat*sign(g_mat[row.names(beta_mat),"Estimate"])
37 |
38 | # Louvain clustering of re-scaled betas
39 |
40 | snn_ref <- BuildSNNSeurat(apply(beta_mat[rowSums(sig_mat != 0) > 1,], 1, function(x){
41 | lower = ifelse(any(x < 0), min(x), 0)
42 | upper = ifelse(any(x > 0), max(x), 0)
43 | extreme = ifelse(abs(lower) > upper, abs(lower), upper)
44 |
45 | return(x/extreme)
46 | }) %>% t, nn.eps = 0)
47 | resolution_list <- c(2,3,4,5)
48 | ids_ref <- Reduce(cbind, mclapply(resolution_list, function(res_use) {
49 | Seurat:::RunModularityClustering(SNN = snn_ref, modularity = 1,
50 | resolution = res_use, algorithm = 1, n.start = 20,
51 | n.iter = 20, random.seed = 100, print.output = FALSE,
52 | temp.file.location = NULL, edge.file.name = NULL)
53 | }, mc.preschedule = FALSE, mc.cores = min(20, length(resolution_list))))
54 | ids_ref <- data.frame(ids_ref)
55 | rm(snn_ref)
56 | gc()
57 |
58 | # Plot heatmap of mean eGene interaction effects in clusters
59 |
60 | pheatmap(sapply(0:7, function(x){colMeans((apply(beta_mat[rowSums(sin_mat != 0) > 1,], 1, function(x){
61 | lower = ifelse(any(x < 0), min(x), 0)
62 | upper = ifelse(any(x > 0), max(x), 0)
63 | extreme = ifelse(abs(lower) > upper, abs(lower), upper)
64 |
65 | return(x/extreme)
66 | }) %>% t)[ids_ref$V2 == x,])}),
67 | color = colorRampPalette(rev(brewer.pal(n = 7, name = "RdBu")))(200),
68 | breaks = seq(-1, 1, .01),
69 | cluster_cols = F, cluster_rows = F, scale = "none", filename = "plot.pdf", height = 6, width = 12)
70 |
71 | # Plot eGene clusters
72 |
73 | fig.size(6,12)
74 | Reduce(`+`, lapply(0:7, function(x){
75 | ggplot(apply(beta_mat[rowSums(sig_mat != 0) > 1,], 1, function(x){
76 | lower = ifelse(any(x < 0), min(x), 0)
77 | upper = ifelse(any(x > 0), max(x), 0)
78 | extreme = ifelse(abs(lower) > upper, abs(lower), upper)
79 |
80 | return(x/extreme)
81 | }) %>% t %>% data.frame %>% filter(ids_ref$V2 == x) %>% rownames_to_column("gene") %>% gather(CV, direction, -gene), aes(x = CV, y = direction, group = gene)) + geom_line(color = "lightgrey") + geom_line(data = data.frame(V1 = paste0("GxCV", 1:7), V2 = colMeans((apply(beta_mat[rowSums(sig_mat != 0) > 1,], 1, function(x){
82 | lower = ifelse(any(x < 0), min(x), 0)
83 | upper = ifelse(any(x > 0), max(x), 0)
84 | extreme = ifelse(abs(lower) > upper, abs(lower), upper)
85 |
86 | return(x/extreme)
87 | }) %>% t)[ids_ref$V2 == x,])), aes(V1, V2, group = NA), color = "red") + theme_classic() + theme(axis.text.x = element_text(angle = 45, hjust = 1, vjust =1
88 | ))
89 | })) + plot_layout(ncol = 4)
90 |
91 | # Test enrichment of GO term gene sets from MSigDB
92 |
93 | gene_sets = msigdbr(species = "Homo sapiens", category = "C5")
94 | msigdbr_list = split(x = gene_sets$gene_symbol, f = gene_sets$gs_name)
95 | enrich_p = lapply(msigdbr_list, function(gene_set){
96 | x = length(intersect(gene_set, row.names(beta_mat[rowSums(sig_mat != 0) > 1,])[ids_ref$V2 == 0]))
97 | y = length(intersect(gene_set, row.names(beta_mat[rowSums(sig_mat != 0) > 1,])[ids_ref$V2 != 0]))
98 |
99 | return(fisher.test(matrix(c(x, y, sum(ids_ref$V2 == 0)-x, sum(ids_ref$V2 != 0)-y), nrow = 2))$p.value)
100 | })
101 | enrich_OR = lapply(msigdbr_list, function(gene_set){
102 | x = length(intersect(gene_set, row.names(beta_mat[rowSums(sig_mat != 0) > 1,])[ids_ref$V2 == 0]))
103 | y = length(intersect(gene_set, row.names(beta_mat[rowSums(sig_mat != 0) > 1,])[ids_ref$V2 != 0]))
104 |
105 | return(fisher.test(matrix(c(x, y, sum(ids_ref$V2 == 0)-x, sum(ids_ref$V2 != 0)-y), nrow = 2))$estimate)
106 | })
107 |
--------------------------------------------------------------------------------
/scripts/post_analysis/enrich_atac.R:
--------------------------------------------------------------------------------
1 | # Script to calculate enrichment of eQTL signal in ATAC-seq peaks from sorted T cell types in Calderon, et al. (2019)Nat Genetics
2 |
3 | # Load libraries
4 | library(data.table)
5 | library(dplyr)
6 | library(tidyr)
7 | library(edgeR)
8 |
9 | args = commandArgs(trailingOnly=TRUE)
10 | celltype = args[1]
11 | n = args[2]
12 |
13 | # Load table of eGenes and lead SNPs
14 |
15 | genes <- read.table("/path/pseudobulk_sig_eqtls.txt", stringsAsFactors = F)
16 |
17 | # Load keys to convert gene names to Ensembl IDs (from Cell Ranger refdata-cellranger-GRCh38-3.0.0) and GRCh38 variants to hg19 (based on liftOver)
18 | ensg_ids <- fread("/path/features.tsv.gz")
19 | conversion_key <- fread("/path/convert_grch38_hg19.txt.gz")
20 |
21 | # Define open regions based on ATAC
22 | atac <- fread("/data/srlab2/anathan/external_data/Calderon2019/GSE118189_ATAC_counts.txt.gz")
23 | atac <- data.frame(atac, check.names = F)
24 | atac <- atac[,grepl(paste(c("V1",unique(substring(colnames(atac), 6, nchar(colnames(atac))-2))[6:19]), collapse = "|"), colnames(atac))]
25 |
26 | atac[,-1] <- cpm(atac[,-1])
27 | atac_bed <- data.frame(V1 = atac$V1) %>% separate(V1, into = c("V1", "V2", "V3"), sep = "_")
28 | atac_bed$V2 <- as.numeric(atac_bed$V2)
29 | atac_bed$V3 <- as.numeric(atac_bed$V3)
30 | atac_bed$V4 = rowMeans(atac[,grepl(paste0("-", celltype, "-U"), colnames(atac))])
31 | atac_bed$size = as.numeric(atac_bed$V3) - as.numeric(atac_bed$V2)
32 |
33 | genes <- genes %>% left_join(ensg_ids, by = c("V3"="V2")) %>% separate(V1, into = c("chr", "pos", "ref", "alt"), sep = "_")
34 |
35 | results <- sapply(1:nrow(genes), function(x) {
36 | if(file.exists(paste0("/path/", genes$chr[x], "/", genes$chr[x], "_", genes$V1.y[x], "_SingleCausal.caviar_post"))) {
37 | cav <- read.table(paste0("/path/", genes$chr[x], "/", genes$chr[x], "_", genes$V2.y[x], "_SingleCausal.caviar_post"), header = T)
38 | if(file.exists(paste0("/path/", genes$chr[x], "/", genes$chr[x], "_", genes$V2.y[x], "_ConditionSecondSNP_SingleCausal.caviar_post"))) {
39 | cav <- read.table(paste0("/path/", genes$chr[x], "/", genes$chr[x], "_", genes$V2.y[x], "_ConditionSecondSNP_SingleCausal.caviar_post"), header = T)}
40 | if(max(cav$Prob_in_pCausalSet) >=.5) {
41 | cav <- cav %>% left_join(conversion_key, by = c("SNP_ID"="V1")) %>% separate(V2, into = c("chr", "pos", "ref", "alt"), sep = "_")
42 | cav$overlap = Infi
43 | tmp <- atac_bed %>% filter(V1 == genes$chr[y])
44 | cav$overlap = sapply(as.numeric(cav$pos), function(x){sum(tmp$V2 <= x & tmp$V3 >= x & tmp$V4 > n)})
45 | e_overlap = sum(cav$overlap)
46 | e_pip = sum(cav$Prob_in_pCausalSet)
47 | e_num = sum(cav$Prob_in_pCausalSet*cav$overlap)
48 | return(c(x, e_overlap, e_pip, e_num, nrow(cav), sum(cumsum(sort(cav$Prob_in_pCausalSet, decreasing = T)) <= .95)))}
49 | else{
50 | return(c(NA,NA,NA,NA,NA,NA))
51 | }}
52 |
53 | else {
54 | return(c(NA,NA,NA,NA,NA,NA))
55 | }})
56 |
57 | # Load eQTL output (data frame from concatenating output of single-cell eQTL models; 195,330 rows = 6,511 eGenes x 30 fixed effects estimated for each)
58 | sceqtl_results <- readRDS("/path/sceqtl_results.rds")
59 |
60 | genes_sig <- sceqtl_results %>% filter(term == "G") %>% mutate(qval = qvalue::qvalue(lrt_pval)$qvalue) %>% filter(qval < .05) %>% select(gene) %>% unlist
61 |
62 | results <- results %>% filter(!is.na(X1))
63 |
64 | # Find intersection with eGenes significant in Peruvian/Blueprint joint meta-analysis
65 | genes_bp <- read.table("/path/joint_bp_sig_eqtls.txt")
66 | genes_int <- genes[genes$V3 %in% genes_bp$V3,]
67 |
68 | # Calculate mean enrichment stats
69 | results %>% mutate(es = X4/X5/(X2/X5*X3/X5)) %>% filter(!is.na(es)) %>% mutate(gene = genes$V3[X1]) %>% filter(gene %in% genes_int$V3) %>% select(es) %>% unlist %>% mean
70 | results %>% mutate(es = X4/X5/(X2/X5*X3/X5)) %>% filter(!is.na(es)) %>% mutate(gene = genes$V3[X1]) %>% filter(gene %in% genes_int$V3) %>% filter(gene %in% genes_sig) %>% select(es) %>% unlist %>% mean
71 | results %>% mutate(es = X4/X5/(X2/X5*X3/X5)) %>% filter(!is.na(es)) %>% mutate(gene = genes$V3[X1]) %>% filter(gene %in% genes_int$V3) %>% filter(!gene %in% genes_sig) %>% select(es) %>% unlist %>% mean
72 |
--------------------------------------------------------------------------------
/scripts/post_analysis/enrich_impact.R:
--------------------------------------------------------------------------------
1 | # Script to calculate enrichment of eQTL signal in T cell-specific regulatory regions defined by IMPACT (Amariuta, et al. [2019] AJHG)
2 |
3 | # Load libraries
4 | library(data.table)
5 | library(dplyr)
6 | library(tidyr)
7 |
8 | # Load table of eGenes and lead SNPs
9 |
10 | genes <- read.table("/path/pseudobulk_sig_eqtls.txt", stringsAsFactors = F)
11 |
12 | # Load keys to convert gene names to Ensembl IDs (from Cell Ranger refdata-cellranger-GRCh38-3.0.0) and GRCh38 variants to hg19 (based on liftOver)
13 | ensg_ids <- fread("/path/features.tsv.gz")
14 | conversion_key <- fread("/path/convert_grch38_hg19.txt.gz")
15 |
16 | # Load +/- 2kb windows around all 6,511 eGenes
17 | tss_track <- read.table("/path/tss_track.txt", header = F)
18 |
19 | genes <- genes %>% left_join(ensg_ids, by = c("V3"="V2")) %>% separate(V1, into = c("chr", "pos", "ref", "alt"), sep = "_")
20 | results <- c()
21 | for(x in 1:22) {
22 | impact <- fread(paste0("/path/IMPACT_predictions_92_chr", x, "_bedgraph.txt.gz"), header = T)
23 |
24 | genes_chr <- genes %>% filter(chr == paste0("chr",x))
25 |
26 | results_chr <- sapply(1:nrow(genes_chr), function(y) {
27 | if(file.exists(paste0("/path/chr", x, "/chr", x, "_", genes_chr$V2.y[y], "_SingleCausal.caviar_post"))) {
28 | cav <- read.table(paste0("/path/chr", x, "/chr", x, "_", genes_chr$V2.y[y], "_SingleCausal.caviar_post"), header = T)
29 | if(file.exists(paste0("/path/chr", x, "/chr", x, "_", genes$V2.y[y], "_ConditionSecondSNP_SingleCausal.caviar_post"))) {
30 | cav <- read.table(paste0("/path/chr", x, "/chr", x, "_", genes$V2.y[y], "_ConditionSecondSNP_SingleCausal.caviar_post"), header = T)}
31 | if(max(cav$Prob_in_pCausalSet) >=.5) {
32 | cav <- cav %>% left_join(conversion_key, by = c("SNP_ID"="V1")) %>% separate(V2, into = c("chr", "pos", "ref", "alt"), sep = "_")
33 | tmp <- impact[tail(which(impact$V2 < min(as.numeric(cav$pos))), n = 1):head(which(impact$V3 > max(as.numeric(cav$pos))), n = 1),]
34 | cav$overlap = Inf
35 | cav$overlap = sapply(as.numeric(cav$pos), function(z){ifelse(any(tss_track$V1 == paste0("chr", x) & tss_track$V2 <= z & tss_track$V3 >= z), 0, tmp$V4[tmp$V2 <= z & tmp$V3 >= z])})
36 | e_overlap = sum(cav$overlap)
37 | e_pip = sum(cav$Prob_in_pCausalSet)
38 | e_num = sum(cav$Prob_in_pCausalSet*cav$overlap)
39 | return(c(which(genes$V2.y == genes_chr$V2.y[y]), e_overlap, e_pip, e_num, nrow(cav), sum(cumsum(sort(cav$Prob_in_pCausalSet, decreasing = T)) <= .95)))}
40 | else{
41 | return(c(NA,NA,NA,NA,NA,NA))
42 | }}
43 |
44 | else {
45 | return(c(NA,NA,NA,NA,NA,NA))
46 | }})
47 | results <- rbind(results, results_chr)
48 | }
49 |
50 | # Load eQTL output (data frame from concatenating output of single-cell eQTL models; 195,330 rows = 6,511 eGenes x 30 fixed effects estimated for each)
51 | sceqtl_results <- readRDS("/path/sceqtl_results.rds")
52 |
53 | genes_sig <- sceqtl_results %>% filter(term == "G") %>% mutate(qval = qvalue::qvalue(lrt_pval)$qvalue) %>% filter(qval < .05) %>% select(gene) %>% unlist
54 |
55 | results <- results %>% filter(!is.na(X1))
56 |
57 | # Find intersection with eGenes significant in Peruvian/Blueprint joint meta-analysis
58 | genes_bp <- read.table("/path/joint_bp_sig_eqtls.txt")
59 | genes_int <- genes[genes$V3 %in% genes_bp$V3,]
60 |
61 | # Calculate mean enrichment stats
62 | results %>% mutate(es = X4/X5/(X2/X5*X3/X5)) %>% filter(!is.na(es)) %>% mutate(gene = genes$V3[X1]) %>% filter(gene %in% genes_int$V3) %>% select(es) %>% unlist %>% mean
63 | results %>% mutate(es = X4/X5/(X2/X5*X3/X5)) %>% filter(!is.na(es)) %>% mutate(gene = genes$V3[X1]) %>% filter(gene %in% genes_int$V3) %>% filter(gene %in% genes_sig) %>% select(es) %>% unlist %>% mean
64 | results %>% mutate(es = X4/X5/(X2/X5*X3/X5)) %>% filter(!is.na(es)) %>% mutate(gene = genes$V3[X1]) %>% filter(gene %in% genes_int$V3) %>% filter(!gene %in% genes_sig) %>% select(es) %>% unlist %>% mean
65 |
--------------------------------------------------------------------------------
/scripts/post_analysis/enrich_tss.R:
--------------------------------------------------------------------------------
1 | # Script to calculate enrichment of eQTL signal in transcription start sites
2 |
3 | # Load libraries
4 | library(data.table)
5 | library(dplyr)
6 | library(tidyr)
7 |
8 | # Load table of eGenes and lead SNPs
9 |
10 | genes <- read.table("/path/pseudobulk_sig_eqtls.txt", stringsAsFactors = F)
11 |
12 | # Load keys to convert gene names to Ensembl IDs (from Cell Ranger refdata-cellranger-GRCh38-3.0.0) and GRCh38 variants to hg19 (based on liftOver)
13 | ensg_ids <- fread("/path/features.tsv.gz")
14 | conversion_key <- fread("/path/convert_grch38_hg19.txt.gz")
15 |
16 | # Load +/- 2kb windows around all 6,511 eGenes
17 | my_tss <- read.table("/path/tss_track.txt", header = F)
18 |
19 | genes <- genes %>% left_join(ensg_ids, by = c("V3"="V2")) %>% separate(V1, into = c("chr", "pos", "ref", "alt"), sep = "_")
20 |
21 | results <- sapply(1:nrow(genes), function(x) {
22 | if(file.exists(paste0("/path/", genes$chr[x], "/", genes$chr[x], "_", genes$V1.y[x], "_SingleCausal.caviar_post"))) {
23 | cav <- read.table(paste0("/path/", genes$chr[x], "/", genes$chr[x], "_", genes$V2.y[x], "_SingleCausal.caviar_post"), header = T)
24 | if(file.exists(paste0("/path/", genes$chr[x], "/", genes$chr[x], "_", genes$V2.y[x], "_ConditionSecondSNP_SingleCausal.caviar_post"))) {
25 | cav <- read.table(paste0("/path/", genes$chr[x], "/", genes$chr[x], "_", genes$V2.y[x], "_ConditionSecondSNP_SingleCausal.caviar_post"), header = T)}
26 | if(max(cav$Prob_in_pCausalSet) >=.5) {
27 | cav <- cav %>% left_join(conversion_key, by = c("SNP_ID"="V1")) %>% separate(V2, into = c("chr", "pos", "ref", "alt"), sep = "_")
28 | cav$overlap = Inf
29 | tmp <- my_tss %>% filter(V1 == genes$chr[x])
30 | cav$overlap = sapply(as.numeric(cav$pos), function(y){sum(tmp$V2 <= y & tmp$V3 >= y)})
31 | e_overlap = sum(cav$overlap)
32 | e_pip = sum(cav$Prob_in_pCausalSet)
33 | e_num = sum(cav$Prob_in_pCausalSet*cav$overlap)
34 | return(c(x, e_overlap, e_pip, e_num, nrow(cav), sum(cumsum(sort(cav$Prob_in_pCausalSet, decreasing = T)) <= .95)))}
35 | else{
36 | return(c(NA,NA,NA,NA,NA,NA))
37 | }}
38 |
39 | else {
40 | return(c(NA,NA,NA,NA,NA,NA))
41 | }})
42 |
43 | # Load eQTL output (data frame from concatenating output of single-cell eQTL models; 195,330 rows = 6,511 eGenes x 30 fixed effects estimated for each)
44 | sceqtl_results <- readRDS("/path/sceqtl_results.rds")
45 |
46 | genes_sig <- sceqtl_results %>% filter(term == "G") %>% mutate(qval = qvalue::qvalue(lrt_pval)$qvalue) %>% filter(qval < .05) %>% select(gene) %>% unlist
47 |
48 | results <- results %>% filter(!is.na(X1))
49 |
50 | # Find intersection with eGenes significant in Peruvian/Blueprint joint meta-analysis
51 | genes_bp <- read.table("/path/joint_bp_sig_eqtls.txt")
52 | genes_int <- genes[genes$V3 %in% genes_bp$V3,]
53 |
54 | # Calculate mean enrichment stats
55 | results %>% mutate(es = X4/X5/(X2/X5*X3/X5)) %>% filter(!is.na(es)) %>% mutate(gene = genes$V3[X1]) %>% filter(gene %in% genes_int$V3) %>% select(es) %>% unlist %>% mean
56 | results %>% mutate(es = X4/X5/(X2/X5*X3/X5)) %>% filter(!is.na(es)) %>% mutate(gene = genes$V3[X1]) %>% filter(gene %in% genes_int$V3) %>% filter(gene %in% genes_sig) %>% select(es) %>% unlist %>% mean
57 | results %>% mutate(es = X4/X5/(X2/X5*X3/X5)) %>% filter(!is.na(es)) %>% mutate(gene = genes$V3[X1]) %>% filter(gene %in% genes_int$V3) %>% filter(!gene %in% genes_sig) %>% select(es) %>% unlist %>% mean
58 |
--------------------------------------------------------------------------------
/scripts/post_analysis/run_coloc.R:
--------------------------------------------------------------------------------
1 | # Script to run coloc on GWAS summary statistics and pseudobulk eQTL results
2 |
3 | # Load libraries
4 |
5 | library(data.table)
6 | library(coloc)
7 | library(dplyr)
8 | library(tidyr)
9 |
10 | # Load table of eGenes and lead SNPs
11 |
12 | genes <- read.table("/path/pseudobulk_sig_eqtls.txt", stringsAsFactors = F)
13 | genes <- genes %>% mutate(snp = V1) %>% separate(V1, into = c("chr", "pos", "ref", "alt"), sep = "_") %>% mutate(chr = gsub("chr", "", chr))
14 | genes$nsnps = NA
15 | genes$PP.H0.abf = NA
16 | genes$PP.H1.abf = NA
17 | genes$PP.H2.abf = NA
18 | genes$PP.H3.abf = NA
19 | genes$PP.H4.abf = NA
20 |
21 | # Load GWAS summary statistics
22 |
23 | gwas_sumstat <- fread("/path/24076602-GCST005531-EFO_0003885.h.tsv.gz")
24 | gwas_sumstat <- gwas_sumstat %>% filter(!is.na(hm_variant_id)) %>% filter(!duplicated(hm_variant_id)) %>% filter(!is.na(hm_beta))
25 |
26 | geno <- fread("/path/geno_imputed_dosage_maf05_hg38_QC.vcf.gz")
27 | meta <- fread("/path/GSE158769_meta_data.txt.gz")
28 |
29 | gwas_dataset = list(type = "cc",
30 | snp = paste0("chr", gwas_sumstat$hm_variant_id),
31 | beta = gwas_sumstat$hm_beta,
32 | varbeta = gwas_sumstat$standard_error^2,
33 | position = gwas_sumstat$hm_pos)
34 | check_dataset(gwas_dataset)
35 |
36 | for(x in 1:22){
37 | eqtl <- fread(paste0("/path/fastqtl_nominal/chr", x, "/chr", x, "_output.nominal.tbru.txt.gz"))
38 |
39 | resid <- data.frame(fread(paste0("/path/chr", x, "_input.bed.gz")))
40 |
41 | for(i in which(genes$chr == x)){
42 | G <- t(subset(geno, ID == sub(":", "_", genes$snp[i]))[,unique(as.character(meta$donor))])[,1]
43 | select_eqtl <- eqtl %>% filter(V1 == genes$V2[i]) %>% mutate(z = qnorm(V4/2, lower.tail = F)*sign(V5), se = V5/z)
44 |
45 | eQTL_dataset = list(beta = select_eqtl$V5,
46 | varbeta = select_eqtl$se^2,
47 | sdY = sd(unlist(resid[resid$ID == genes$V2[i],-c(1:4)])),
48 | type = "quant",
49 | snp = select_eqtl$V2)
50 | tryCatch({
51 | my.res <- coloc.abf(dataset1=eQTL_dataset,
52 | dataset2=gwas_dataset)
53 |
54 | genes[i,8:13] = my.res$summary
55 | }, error=function(cond){return(NA)})
56 | }
57 | }
58 |
--------------------------------------------------------------------------------
/scripts/post_analysis/run_homer.sh:
--------------------------------------------------------------------------------
1 | #!/bin/bash
2 |
3 | # Script to run HOMER on promoters of eGenes interacting with given CV
4 | x=$1
5 |
6 | perl /path/homer/bin/findMotifs.pl /path/homer_out/CV${x}_genes.txt human /path/homer_out/CV${x}/ -start -2000 -end 2000 -len 8,10 -p 4
7 |
8 | # Script to run HOMER on BED file of +/- 20bp windows around eQTL variants interacting with given CV
9 | x=$1
10 |
11 | perl /path/homer/bin/findMotifsGenome.pl /path/homer_out/CV${x}_snps.bed hg19 /path/homer_out/CV${x}_snps_given/ -size given
12 |
--------------------------------------------------------------------------------
/scripts/pseudobulk/01_make_pseudobulk_exprs.R:
--------------------------------------------------------------------------------
1 | # Script to make pseudobulk expression profiles
2 |
3 | # Load libraries
4 | library(data.table)
5 | library(presto)
6 | library(DESeq2)
7 | library(edgeR)
8 | library(dplyr)
9 |
10 | # Load metadata (from GEO: GSE158769)
11 | meta_data <- fread("/path/GSE158769_meta_data.txt.gz")
12 | meta_data <- meta_data %>% mutate(id = paste(donor, batch, sep = "_"))
13 |
14 | # Load RAW UMI COUNTS matrix (from GEO)
15 | exprs_raw <- fread("/path/GSE158769_exprs_raw.tsv.gz")
16 |
17 | # Remove replicates
18 | reps <- meta_data %>% filter(!duplicated(id)) %>% arrange(batch) %>% filter(duplicated(donor)) %>% select(id) %>% unlist
19 | meta_data <- meta_data %>% filter(!id %in% reps)
20 | exprs_raw <- exprs_raw[,meta_data$cell_id]
21 |
22 | # Make pseudobulk profiles; specify variable to make pseudobulk samples for
23 | all_collapse <- collapse_counts(exprs_raw, meta_data, c("donor"))
24 | colnames(all_collapse$counts_mat) <- all_collapse$meta_data$donor
25 |
26 | # Save genes (rows) x samples (columns) matrix
27 |
28 | out <- data.frame(id=row.names(all_collapse$counts_mat), all_collapse$counts_mat)
29 |
30 | gz1 <- gzfile("/path/pseudobulk_exprs_mat.txt.gz","w")
31 | write.table(out, gz1, sep = "\t", quote = F, row.names = FALSE)
32 | close(gz1)
33 |
--------------------------------------------------------------------------------
/scripts/pseudobulk/02_normalize.R:
--------------------------------------------------------------------------------
1 | # Script to normalize pseudobulk expression
2 |
3 | # Load libraries
4 | library(ggplot2)
5 | library(Seurat)
6 | library(data.table)
7 | library(ggrepel)
8 | library(peer)
9 | library(edgeR)
10 |
11 | # Load donor-level covariates (from dbGaP: phs002467)
12 | pheno <- read.delim("/path/donor_covariates.txt", sep = " ")
13 | pheno <- pheno[!duplicated(pheno$donor),]
14 | row.names(pheno) <- pheno$donor
15 |
16 | # Load genotype PCs
17 | pcs <- read.delim("/path/geno_PCs.txt", sep = " ", header = F)
18 | row.names(pcs) <- pcs$V2
19 | pcs <- pcs[,3:ncol(pcs)]
20 |
21 | # PEER normalization: regress out latent variables (Stegle 2010, PLoS Comp Bio)
22 |
23 | # Based on GTEx 2017 Nature:
24 | K = 45
25 |
26 | # Replace with reading in your pseudobulk expression matrix
27 | sum_exp <- fread("/path/pseudobulk_exprs_mat.txt.gz")
28 | sum_exp <- as.data.frame(sum_exp)
29 | row.names(sum_exp) <- sum_exp[,1]
30 | sum_exp <- sum_exp[,-1]
31 |
32 | # Remove genes that with non-zero expression in <= half of the samples
33 | sum_exp <- sum_exp[rowSums(sum_exp > 0) > .5*ncol(sum_exp),]
34 |
35 | # Log2 CPM normalization
36 | norm_exp <- log2(cpm(sum_exp)+1)
37 |
38 | # Inverse normal transformation
39 | rn<-apply(norm_exp,1,function(x){
40 | qnorm( (rank(x, na.last="keep") - 0.5) / sum(!is.na(x)) )
41 | })
42 | rn<-t(rn)
43 |
44 | # Select known covariates to regress out (age, sex, and 5 genotype PCs)
45 | covs <- cbind(pheno[colnames(norm_exp),"age"],
46 | pheno[colnames(norm_exp),"Sex"] == "F",
47 | pcs[colnames(norm_exp),1:5])
48 | colnames(covs) <- c("age", "female", "PC1", "PC2", "PC3", "PC4", "PC5")
49 |
50 | model = PEER()
51 | PEER_setPhenoMean(model, as.matrix(t(rn)))
52 | PEER_setAdd_mean(model, TRUE)
53 | PEER_setNk(model,K)
54 | PEER_getNk(model)
55 | PEER_setCovariates(model, as.matrix(covs))
56 | PEER_setNmax_iterations(model,10000)
57 | PEER_update(model)
58 |
59 | # Save residuals
60 | residuals = PEER_getResiduals(model)
61 | colnames(residuals) <- row.names(norm_exp)
62 | row.names(residuals) <- colnames(norm_exp)
63 | dump <- data.frame( ids = row.names(norm_exp),t(residuals))
64 | gz1 <- gzfile("/path/peer_residuals.txt.gz","w")
65 | write.table(dump, gz1, sep = "\t", quote = F, row.names = FALSE)
66 | close(gz1)
67 |
--------------------------------------------------------------------------------
/scripts/pseudobulk/03_make_BED.sh:
--------------------------------------------------------------------------------
1 | #!/bin/bash
2 |
3 | # Script to make BED file from PEER residuals
4 |
5 | # BED header
6 | echo -e "#Chr\tStart\tEnd\tID" > tmp
7 | zcat /path/peer_residuals.txt.gz | head -n1 | cut -f2- | sed 's/\./-/g' | paste tmp - > header
8 |
9 | zcat /path/peer_residuals.txt.gz | sed -e "1d" |
10 | sort -k 1b,1 > tmp01
11 |
12 | # Needs to be by chromosome for FastQTL
13 | for chr in $(seq 1 22);do
14 | echo chr$chr ...
15 |
16 | # Get gene TSS using GTFs from Cell Ranger (refdata-cellranger-GRCh38-3.0.0)
17 | sed -e "1d" /path/chr${chr}_tss_gene_basic.bed |
18 | awk -v chr=$chr '{if($1==chr){print $2,chr,$4 -1,$4}}' |
19 | awk '{ $1 = substr($1, 1, 15)} 1' |
20 | sort -k 1b,1 > tmp02
21 |
22 | outfile="/path/chr${chr}_input.bed"
23 | # Link ENSG to gene name using features.tsv.gz file from cellranger-3.1.0, GRCh38
24 | zcat /path/features.tsv.gz |
25 | grep ^ENSG |
26 | sort -k 1b,1 |
27 | join - tmp02 |
28 | awk '{print $2,$5,$6,$7,$2}' |
29 | sort -k 1b,1 |
30 | uniq -f 4 |
31 | join - tmp01 |
32 | cut -d " " -f2- |
33 | sort -k 2,2n |
34 | perl -pe "s/ /\t/g" |
35 | awk '{print $0}' |
36 | cat header - > "$outfile"
37 |
38 | bgzip -f "$outfile"
39 |
40 | tabix -f -p bed "$outfile.gz"
41 | done
42 |
--------------------------------------------------------------------------------
/scripts/pseudobulk/04_fastqtl_nominal.sh:
--------------------------------------------------------------------------------
1 | #!/bin/sh
2 |
3 | # Script to run FastQTL (Ongen 2015, Bioinformatics)
4 |
5 | # VCF: post QC and imputation genotype dosage VCF (genotype data from dbGaP: phs002025)
6 | # BED: output from previous script
7 | # Region: select chromosome
8 | # out: output file
9 |
10 | chr=$1
11 |
12 | mkdir -p /path/fastqtl_nominal/chr$chr
13 |
14 | /path/fastQTL \
15 | --vcf /path/geno_imputed_dosage_maf05_hg38_QC.vcf.gz \
16 | --bed /path/chr${chr}_input.bed.gz \
17 | --region chr$chr \
18 | --out /path/fastqtl_nominal/chr$chr/chr${chr}_output.nominal.tbru.txt
19 |
20 | gzip -f /path/fastqtl_nominal/chr$chr/chr${chr}_output.nominal.tbru.txt
21 |
--------------------------------------------------------------------------------
/scripts/pseudobulk/05_fastqtl_permute.sh:
--------------------------------------------------------------------------------
1 | #!/bin/sh
2 |
3 | # Script to run FastQTL (Ongen 2015, Bioinformatics)
4 |
5 | # VCF: post QC and imputation genotype dosage VCF (genotype data from dbGaP: phs002025)
6 | # BED: output from previous script
7 | # Region: select chromosome
8 | # out: output file
9 |
10 | chr=$1
11 |
12 | mkdir -p /path/fastqtl_permute/chr$chr
13 |
14 | /path/fastQTL \
15 | --vcf /path/geno_imputed_dosage_maf05_hg38_QC.vcf.gz \
16 | --bed /path/chr${chr}_input.bed.gz \
17 | --region chr$chr \
18 | --permute 1000 \
19 | --out /path/fastqtl_nominal/chr$chr/chr${chr}_output.permute.tbru.txt
20 |
21 | gzip -f /path/fastqtl_nominal/chr$chr/chr${chr}_output.permute.tbru.txt
22 |
--------------------------------------------------------------------------------
/scripts/singlecell/linear_nostate.R:
--------------------------------------------------------------------------------
1 | # Script to run single-cell linear mixed effects eQTL model without cell state interaction
2 |
3 | # Load libraries
4 | library(lme4)
5 | library(Matrix)
6 |
7 | args = commandArgs(trailingOnly=TRUE)
8 |
9 | gene <- args[1]
10 | snp <- args[2]
11 |
12 | # Load normalized UMI counts matrix and metadata (from GEO: GSE158769) and gene expression PCs
13 | exprs_norm <- fread("/path/GSE158769_exprs_norm.tsv.gz")
14 | meta <- fread("/path/GSE158769_meta_data.txt.gz")
15 | pca_res <- readRDS("/path/pca_irlba.rds")
16 | pca_res <- pca_res$x
17 |
18 | # Load genotype dosages (from dbGaP: phs002025) and PCs
19 | geno <- fread("/path/geno_imputed_dosage_maf05_hg38_QC.vcf.gz")
20 | pcs <- read.delim("/path/geno_PCs.txt", sep = " ", header = F)
21 | row.names(pcs) <- pcs$V2
22 | pcs <- pcs[,3:ncol(pcs)]
23 |
24 | # Make data frame of variables for model
25 | E <- as.numeric(exprs_norm[gene,])
26 | G <- t(subset(geno, ID == snp)[,as.character(meta$donor)])[,1]
27 | IND <- factor(meta$donor)
28 | B <- meta$batch
29 | AGE <- scale(meta$tbru_age)
30 | SEX <- meta$Sex
31 | nUMI <- scale(log(meta$nUMI))
32 | MT <- meta$percent_mito
33 | PC <- pcs[as.character(meta$donor),1:5]
34 | expPC <- pca_res[,1:5]
35 |
36 | data <- data.frame(E,G,IND,B,AGE,SEX,nUMI,MT,
37 | PC1 = PC[,1],PC2 = PC[,2],PC3 = PC[,3],PC4 = PC[,4],PC5 = PC[,5],
38 | expPC1 = expPC[,1], expPC2 = expPC[,2], expPC3 = expPC[,3], expPC4 = expPC[,4], expPC5 = expPC[,5])
39 | data$G <- as.numeric(as.character(data$G))
40 |
41 | full_model <- lme4::lmer(formula = E ~ G + (1 | IND) + (1 | B) + AGE + SEX + nUMI + MT + PC1 + PC2 + PC3 +PC4 + PC5 + expPC1 + expPC2 + expPC3 + expPC4 + expPC5,
42 | data= data, REML = F)
43 | null_model <- lme4::lmer(formula = E ~ (1 | IND) + (1 | B) + AGE + SEX + nUMI + MT + PC1 + PC2 + PC3 + PC4 + PC5 + expPC1 + expPC2 + expPC3 + expPC4 + expPC5,
44 | data= data, REML = F)
45 | model_lrt <- anova(null_model, full_model)
46 |
47 | out <- summary(test)$coefficients
48 | colnames(out) <- c("Estimate","Std.Error","tvalue")
49 | out <- data.frame(gene=gene,snp=snp,term=row.names(x),x, lrt_pval=model_lrt$`Pr(>Chisq)`[2])
50 |
--------------------------------------------------------------------------------
/scripts/singlecell/linear_nostate_permute.R:
--------------------------------------------------------------------------------
1 | # Script to run single-cell linear mixed effects eQTL model without cell state interaction on permuted data (permuting genotype)
2 |
3 | # Load libraries
4 | library(lme4)
5 | library(Matrix)
6 |
7 | args = commandArgs(trailingOnly=TRUE)
8 |
9 | gene <- args[1]
10 | snp <- args[2]
11 |
12 | # Load normalized UMI counts matrix and metadata (from GEO: GSE158769) and gene expression PCs
13 | exprs_norm <- fread("/path/GSE158769_exprs_norm.tsv.gz")
14 | meta <- fread("/path/GSE158769_meta_data.txt.gz")
15 | pca_res <- readRDS("/path/pca_irlba.rds")
16 | pca_res <- pca_res$x
17 |
18 | # Load genotype dosages (from dbGaP: phs002025) and PCs
19 | geno <- fread("/path/geno_imputed_dosage_maf05_hg38_QC.vcf.gz")
20 | pcs <- read.delim("/path/geno_PCs.txt", sep = " ", header = F)
21 | row.names(pcs) <- pcs$V2
22 | pcs <- pcs[,3:ncol(pcs)]
23 |
24 | # Make data frame of variables for model
25 | set.seed(1010*as.numeric)
26 | E <- as.numeric(exprs_norm[gene,])
27 | G <- subset(geno, ID == snp)
28 | colnames(G)[10:276] <- sample(colnames(G)[10:276])
29 | G <- t(G[,as.character(meta$donor)])[,1]
30 | IND <- factor(meta$donor)
31 | B <- meta$batch
32 | AGE <- scale(meta$tbru_age)
33 | SEX <- meta$Sex
34 | nUMI <- scale(log(meta$nUMI))
35 | MT <- meta$percent_mito
36 | PC <- pcs[as.character(meta$donor),1:5]
37 | expPC <- pca_res[,1:5]
38 |
39 | data <- data.frame(E,G,IND,B,AGE,SEX,nUMI,MT,
40 | PC1 = PC[,1],PC2 = PC[,2],PC3 = PC[,3],PC4 = PC[,4],PC5 = PC[,5],
41 | expPC1 = expPC[,1], expPC2 = expPC[,2], expPC3 = expPC[,3], expPC4 = expPC[,4], expPC5 = expPC[,5])
42 | data$G <- as.numeric(as.character(data$G))
43 |
44 | full_model <- lme4::lmer(formula = E ~ G + (1 | IND) + (1 | B) + AGE + SEX + nUMI + MT + PC1 + PC2 + PC3 +PC4 + PC5 + expPC1 + expPC2 + expPC3 + expPC4 + expPC5,
45 | data= data, REML = F)
46 | null_model <- lme4::lmer(formula = E ~ (1 | IND) + (1 | B) + AGE + SEX + nUMI + MT + PC1 + PC2 + PC3 + PC4 + PC5 + expPC1 + expPC2 + expPC3 + expPC4 + expPC5,
47 | data= data, REML = F)
48 | model_lrt <- anova(null_model, full_model)
49 |
50 | out <- summary(test)$coefficients
51 | colnames(out) <- c("Estimate","Std.Error","tvalue")
52 | out <- data.frame(gene=gene,snp=snp,term=row.names(x),x, lrt_pval=model_lrt$`Pr(>Chisq)`[2])
53 |
--------------------------------------------------------------------------------
/scripts/singlecell/linear_univariate.R:
--------------------------------------------------------------------------------
1 | # Script to run single-cell linear mixed effects eQTL model with one cell state interaction (e.g., CV, PC, CD4+, trajectory, disease status)
2 |
3 | # Load libraries
4 | library(lme4)
5 | library(Matrix)
6 |
7 | args = commandArgs(trailingOnly=TRUE)
8 |
9 | gene <- args[1]
10 | snp <- args[2]
11 |
12 | # Load normalized UMI counts matrix and metadata (from GEO: GSE158769) and gene expression PCs
13 | exprs_norm <- fread("/path/GSE158769_exprs_norm.tsv.gz")
14 | meta <- fread("/path/GSE158769_meta_data.txt.gz")
15 | pca_res <- readRDS("/path/pca_irlba.rds")
16 | pca_res <- pca_res$x
17 |
18 | # Load genotype dosages (from dbGaP: phs002025) and PCs
19 | geno <- fread("/path/geno_imputed_dosage_maf05_hg38_QC.vcf.gz")
20 | pcs <- read.delim("/path/geno_PCs.txt", sep = " ", header = F)
21 | row.names(pcs) <- pcs$V2
22 | pcs <- pcs[,3:ncol(pcs)]
23 |
24 | # Make data frame of variables for model
25 | E <- as.numeric(exprs_norm[gene,])
26 | G <- t(subset(geno, ID == snp)[,as.character(meta$donor)])[,1]
27 | IND <- factor(meta$donor)
28 | B <- meta$batch
29 | AGE <- scale(meta$tbru_age)
30 | SEX <- meta$Sex
31 | nUMI <- scale(log(meta$nUMI))
32 | MT <- meta$percent_mito
33 | PC <- pcs[as.character(meta$donor),1:5]
34 | expPC <- pca_res[,1:5]
35 |
36 | # Load interaction covariate (e.g., CD4+ binary annotation from gating on CITE-seq surface protein)
37 | cov <- readRDS("/path/citeseq_gates.rds")[,"cd4"]
38 |
39 | data <- data.frame(E,G,IND,B,AGE,SEX,nUMI,MT,
40 | PC1 = PC[,1],PC2 = PC[,2],PC3 = PC[,3],PC4 = PC[,4],PC5 = PC[,5],
41 | expPC1 = expPC[,1], expPC2 = expPC[,2], expPC3 = expPC[,3], expPC4 = expPC[,4], expPC5 = expPC[,5], cov)
42 | data$G <- as.numeric(as.character(data$G))
43 |
44 | full_model <- lme4::lmer(formula = E ~ G + (1 | B) + (1 | IND) + AGE + SEX + nUMI + MT + PC1 + PC2 + PC3 + PC4 + PC5 + expPC1 + expPC2 + expPC3 + expPC4 + expPC5 + cov + G*cov,
45 | data= data, REML = F)
46 | null_model <- lme4::lmer(formula = E ~ G + (1 | B) + (1 | IND) + AGE + SEX + nUMI + MT + PC1 + PC2 + PC3 + PC4 + PC5 + expPC1 + expPC2 + expPC3 + expPC4 + expPC5 + cov,
47 | data= data, REML = F)
48 | model_lrt <- anova(null_model, full_model)
49 |
50 | out <- summary(test)$coefficients
51 | colnames(out) <- c("Estimate","Std.Error","tvalue")
52 | out <- data.frame(gene=gene,snp=snp,term=row.names(x),x, lrt_pval=model_lrt$`Pr(>Chisq)`[2])
53 |
--------------------------------------------------------------------------------
/scripts/singlecell/linear_univariate_permute.R:
--------------------------------------------------------------------------------
1 | # Script to run single-cell linear mixed effects eQTL model with one cell state interaction (e.g., CV, PC, CD4+, trajectory, disease status) on permuted data (permuting cell state)
2 |
3 | # Load libraries
4 | library(lme4)
5 | library(Matrix)
6 |
7 | args = commandArgs(trailingOnly=TRUE)
8 |
9 | gene <- args[1]
10 | snp <- args[2]
11 |
12 | # Load normalized UMI counts matrix and metadata (from GEO: GSE158769) and gene expression PCs
13 | exprs_norm <- fread("/path/GSE158769_exprs_norm.tsv.gz")
14 | meta <- fread("/path/GSE158769_meta_data.txt.gz")
15 | pca_res <- readRDS("/path/pca_irlba.rds")
16 | pca_res <- pca_res$x
17 |
18 | # Load genotype dosages (from dbGaP: phs002025) and PCs
19 | geno <- fread("/path/geno_imputed_dosage_maf05_hg38_QC.vcf.gz")
20 | pcs <- read.delim("/path/geno_PCs.txt", sep = " ", header = F)
21 | row.names(pcs) <- pcs$V2
22 | pcs <- pcs[,3:ncol(pcs)]
23 |
24 | # Load interaction covariate (e.g., CD4+ binary annotation from gating on CITE-seq surface protein)
25 | set.seed(1010)
26 | cov <- unlist(readRDS("/path/citeseq_gates.rds")[,"cd4"])
27 | for(x in unique(meta$donor)){
28 | cov[meta$donor == x] <- sample(cov[meta$donor == x])
29 | }
30 |
31 | # Make data frame of variables for model
32 | E <- as.numeric(exprs_norm[gene,])
33 | G <- t(subset(geno, ID == snp)[,as.character(meta$donor)])[,1]
34 | IND <- factor(meta$donor)
35 | B <- meta$batch
36 | AGE <- scale(meta$tbru_age)
37 | SEX <- meta$Sex
38 | nUMI <- scale(log(meta$nUMI))
39 | MT <- meta$percent_mito
40 | PC <- pcs[as.character(meta$donor),1:5]
41 | expPC <- pca_res[,1:5]
42 |
43 | data <- data.frame(E,G,IND,B,AGE,SEX,nUMI,MT,
44 | PC1 = PC[,1],PC2 = PC[,2],PC3 = PC[,3],PC4 = PC[,4],PC5 = PC[,5],
45 | expPC1 = expPC[,1], expPC2 = expPC[,2], expPC3 = expPC[,3], expPC4 = expPC[,4], expPC5 = expPC[,5], cov)
46 | data$G <- as.numeric(as.character(data$G))
47 |
48 | full_model <- lme4::lmer(formula = E ~ G + (1 | B) + (1 | IND) + AGE + SEX + nUMI + MT + PC1 + PC2 + PC3 + PC4 + PC5 + expPC1 + expPC2 + expPC3 + expPC4 + expPC5 + cov + G*cov,
49 | data= data, REML = F)
50 | null_model <- lme4::lmer(formula = E ~ G + (1 | B) + (1 | IND) + AGE + SEX + nUMI + MT + PC1 + PC2 + PC3 + PC4 + PC5 + expPC1 + expPC2 + expPC3 + expPC4 + expPC5 + cov,
51 | data= data, REML = F)
52 | model_lrt <- anova(null_model, full_model)
53 |
54 | out <- summary(test)$coefficients
55 | colnames(out) <- c("Estimate","Std.Error","tvalue")
56 | out <- data.frame(gene=gene,snp=snp,term=row.names(x),x, lrt_pval=model_lrt$`Pr(>Chisq)`[2])
57 |
--------------------------------------------------------------------------------
/scripts/singlecell/linear_univariate_simulDE.R:
--------------------------------------------------------------------------------
1 | # Script to run single-cell linear mixed effects eQTL model with one cell state interaction (CD4+), simulating differential expression
2 |
3 | # Load libraries
4 | library(lme4)
5 | library(Matrix)
6 |
7 | args = commandArgs(trailingOnly=TRUE)
8 |
9 | gene <- args[1]
10 | snp <- args[2]
11 | frac <- as.numeric(args[3]) # fraction to which to decrease expression in CD4+ cells
12 |
13 | # Load normalized UMI counts matrix and metadata (from GEO: GSE158769) and gene expression PCs
14 | exprs_norm <- fread("/path/GSE158769_exprs_norm.tsv.gz")
15 | meta <- fread("/path/GSE158769_meta_data.txt.gz")
16 | pca_res <- readRDS("/path/pca_irlba.rds")
17 | pca_res <- pca_res$x
18 |
19 | # Load genotype dosages (from dbGaP: phs002025) and PCs
20 | geno <- fread("/path/geno_imputed_dosage_maf05_hg38_QC.vcf.gz")
21 | pcs <- read.delim("/path/geno_PCs.txt", sep = " ", header = F)
22 | row.names(pcs) <- pcs$V2
23 | pcs <- pcs[,3:ncol(pcs)]
24 |
25 | # Make data frame of variables for model
26 | E <- as.numeric(exprs_norm[gene,])
27 | G <- t(subset(geno, ID == snp)[,as.character(meta$donor)])[,1]
28 | IND <- factor(meta$donor)
29 | B <- meta$batch
30 | AGE <- scale(meta$tbru_age)
31 | SEX <- meta$Sex
32 | nUMI <- scale(log(meta$nUMI))
33 | MT <- meta$percent_mito
34 | PC <- pcs[as.character(meta$donor),1:5]
35 | expPC <- pca_res[,1:5]
36 |
37 | # Load interaction covariate (e.g., CD4+ binary annotation from gating on CITE-seq surface protein)
38 | cov <- readRDS("/path/citeseq_gates.rds")[,"cd4"]
39 |
40 | data <- data.frame(E,G,IND,B,AGE,SEX,nUMI,MT,
41 | PC1 = PC[,1],PC2 = PC[,2],PC3 = PC[,3],PC4 = PC[,4],PC5 = PC[,5],
42 | expPC1 = expPC[,1], expPC2 = expPC[,2], expPC3 = expPC[,3], expPC4 = expPC[,4], expPC5 = expPC[,5], cov)
43 | data$G <- as.numeric(as.character(data$G))
44 | data$E[data$cov == TRUE] <- data$E[data$cov == TRUE]/frac
45 |
46 | full_model <- lme4::lmer(formula = E ~ G + (1 | B) + (1 | IND) + AGE + SEX + nUMI + MT + PC1 + PC2 + PC3 + PC4 + PC5 + expPC1 + expPC2 + expPC3 + expPC4 + expPC5 + cov + G*cov,
47 | data= data, REML = F)
48 | null_model <- lme4::lmer(formula = E ~ G + (1 | B) + (1 | IND) + AGE + SEX + nUMI + MT + PC1 + PC2 + PC3 + PC4 + PC5 + expPC1 + expPC2 + expPC3 + expPC4 + expPC5 + cov,
49 | data= data, REML = F)
50 | model_lrt <- anova(null_model, full_model)
51 |
52 | out <- summary(test)$coefficients
53 | colnames(out) <- c("Estimate","Std.Error","tvalue")
54 | out <- data.frame(gene=gene,snp=snp,term=row.names(x),x, lrt_pval=model_lrt$`Pr(>Chisq)`[2])
55 |
--------------------------------------------------------------------------------
/scripts/singlecell/nb_univariate.R:
--------------------------------------------------------------------------------
1 | # Script to run single-cell negative binomial mixed effects eQTL model with one cell state interaction (e.g., CV, PC, CD4+, trajectory, disease status)
2 |
3 | # Load libraries
4 | library(lme4)
5 | library(Matrix)
6 |
7 | args = commandArgs(trailingOnly=TRUE)
8 |
9 | gene <- args[1]
10 | snp <- args[2]
11 |
12 | # Load raw UMI counts matrix and metadata (from GEO: GSE158769) and gene expression PCs
13 | exprs_raw <- fread("/path/GSE158769_exprs_raw.tsv.gz")
14 | meta <- fread("/path/GSE158769_meta_data.txt.gz")
15 | pca_res <- readRDS("/path/pca_irlba.rds")
16 | pca_res <- pca_res$x
17 |
18 | # Load genotype dosages (from dbGaP: phs002025) and PCs
19 | geno <- fread("/path/geno_imputed_dosage_maf05_hg38_QC.vcf.gz")
20 | pcs <- read.delim("/path/geno_PCs.txt", sep = " ", header = F)
21 | row.names(pcs) <- pcs$V2
22 | pcs <- pcs[,3:ncol(pcs)]
23 |
24 | # Make data frame of variables for model
25 | E <- as.numeric(exprs_raw[gene,])
26 | G <- t(subset(geno, ID == snp)[,as.character(meta$donor)])[,1]
27 | IND <- factor(meta$donor)
28 | B <- meta$batch
29 | AGE <- scale(meta$tbru_age)
30 | SEX <- meta$Sex
31 | nUMI <- scale(log(meta$nUMI))
32 | MT <- meta$percent_mito
33 | PC <- pcs[as.character(meta$donor),1:5]
34 | expPC <- pca_res[,1:5]
35 |
36 | # Load interaction covariate (e.g., batch-corrected CV from CCA)
37 | cov <- readRDS("/path/cca_donor_batch.rds")[,1]
38 |
39 | data <- data.frame(E,G,IND,B,AGE,SEX,nUMI,MT,
40 | PC1 = PC[,1],PC2 = PC[,2],PC3 = PC[,3],PC4 = PC[,4],PC5 = PC[,5],
41 | expPC1 = expPC[,1], expPC2 = expPC[,2], expPC3 = expPC[,3], expPC4 = expPC[,4], expPC5 = expPC[,5], cov)
42 | data$G <- as.numeric(as.character(data$G))
43 |
44 | full_model <- lme4::glmer.nb(formula = E ~ G + (1 | B) + (1 | IND) + AGE + SEX + nUMI + MT + PC1 + PC2 + PC3 + PC4 + PC5 + expPC1 + expPC2 + expPC3 + expPC4 + expPC5 + cov + G:cov,
45 | data= data, nAGQ = 0, control = glmerControl(optimizer = "nloptwrap"))
46 | null_model <- lme4::glmer.nb(formula = E ~ G + (1 | B) + (1 | IND) + AGE + SEX + nUMI + MT + PC1 + PC2 + PC3 + PC4 + PC5 + expPC1 + expPC2 + expPC3 + expPC4 + expPC5 + cov,
47 | data= data, nAGQ = 0, control = glmerControl(optimizer = "nloptwrap"))
48 | model_lrt <- anova(null_model, full_model)
49 |
50 | out <- summary(test)$coefficients
51 | colnames(out) <- c("Estimate","Std.Error","zvalue","pval")
52 | out <- data.frame(gene=gene,snp=snp,term=row.names(x),x, lrt_pval=model_lrt$`Pr(>Chisq)`[2])
53 |
--------------------------------------------------------------------------------
/scripts/singlecell/poisson_multivariate.R:
--------------------------------------------------------------------------------
1 | # Script to run single-cell Poisson mixed effects eQTL model with multiple cell state interactions (e.g., CVs, PCs)
2 |
3 | # Load libraries
4 | library(lme4)
5 | library(Matrix)
6 |
7 | args = commandArgs(trailingOnly=TRUE)
8 |
9 | gene <- args[1]
10 | snp <- args[2]
11 |
12 | # Load raw UMI counts matrix and metadata (from GEO: GSE158769) and gene expression PCs
13 | exprs_raw <- fread("/path/GSE158769_exprs_raw.tsv.gz")
14 | meta <- fread("/path/GSE158769_meta_data.txt.gz")
15 | pca_res <- readRDS("/path/pca_irlba.rds")
16 | pca_res <- pca_res$x
17 |
18 | # Load genotype dosages (from dbGaP: phs002025) and PCs
19 | geno <- fread("/path/geno_imputed_dosage_maf05_hg38_QC.vcf.gz")
20 | pcs <- read.delim("/path/geno_PCs.txt", sep = " ", header = F)
21 | row.names(pcs) <- pcs$V2
22 | pcs <- pcs[,3:ncol(pcs)]
23 |
24 | # Make data frame of variables for model
25 | E <- as.numeric(exprs_raw[gene,])
26 | G <- t(subset(geno, ID == snp)[,as.character(meta$donor)])[,1]
27 | IND <- factor(meta$donor)
28 | B <- meta$batch
29 | AGE <- scale(meta$tbru_age)
30 | SEX <- meta$Sex
31 | nUMI <- scale(log(meta$nUMI))
32 | MT <- meta$percent_mito
33 | PC <- pcs[as.character(meta$donor),1:5]
34 | expPC <- pca_res[,1:5]
35 |
36 | # Load interaction covariates (e.g., batch-corrected CVs from CCA)
37 | cov <- readRDS("/path/cca_donor_batch.rds")[,1:7]
38 |
39 | data <- data.frame(E,G,IND,B,AGE,SEX,nUMI,MT,
40 | PC1 = PC[,1],PC2 = PC[,2],PC3 = PC[,3],PC4 = PC[,4],PC5 = PC[,5],
41 | expPC1 = expPC[,1], expPC2 = expPC[,2], expPC3 = expPC[,3], expPC4 = expPC[,4], expPC5 = expPC[,5], cov)
42 | data$G <- as.numeric(as.character(data$G))
43 |
44 | full_model <- lme4::glmer(formula = paste0("E ~ G + (1 | B) + (1 | IND) + AGE + SEX + nUMI + MT + PC1 + PC2 + PC3 + PC4 + PC5 + expPC1 + expPC2 + expPC3 + expPC4 + expPC5 + ",
45 | paste0("CV", 1:7, collapse = " + ")," + ", paste0("G:CV", 1:7, collapse = " + ")),
46 | family = "poisson", nAGQ = 0, data= data, control = glmerControl(optimizer = "nloptwrap"))
47 | null_model <- lme4::glmer(formula = paste0("E ~ G + (1 | B) + (1 | IND) + AGE + SEX + nUMI + MT + PC1 + PC2 + PC3 + PC4 + PC5 + expPC1 + expPC2 + expPC3 + expPC4 + expPC5 + ",
48 | paste0("CV", 1:7, collapse = " + ")),
49 | family = "poisson", nAGQ = 0, data= data, control = glmerControl(optimizer = "nloptwrap"))
50 | model_lrt <- anova(null_model, full_model)
51 |
52 | out <- summary(test)$coefficients
53 | colnames(out) <- c("Estimate","Std.Error","zvalue","pval")
54 | out <- data.frame(gene=gene,snp=snp,term=row.names(out),out, lrt_pval=model_lrt$`Pr(>Chisq)`[2])
55 |
--------------------------------------------------------------------------------
/scripts/singlecell/poisson_nostate.R:
--------------------------------------------------------------------------------
1 | # Script to run single-cell Poisson mixed effects eQTL model without cell state interaction
2 |
3 | # Load libraries
4 | library(lme4)
5 | library(Matrix)
6 |
7 | args = commandArgs(trailingOnly=TRUE)
8 |
9 | gene <- args[1]
10 | snp <- args[2]
11 |
12 | # Load raw UMI counts matrix and metadata (from GEO: GSE158769) and gene expression PCs
13 | exprs_raw <- fread("/path/GSE158769_exprs_raw.tsv.gz")
14 | meta <- fread("/path/GSE158769_meta_data.txt.gz")
15 | pca_res <- readRDS("/path/pca_irlba.rds")
16 | pca_res <- pca_res$x
17 |
18 | # Load genotype dosages (from dbGaP: phs002025) and PCs
19 | geno <- fread("/path/geno_imputed_dosage_maf05_hg38_QC.vcf.gz")
20 | pcs <- read.delim("/path/geno_PCs.txt", sep = " ", header = F)
21 | row.names(pcs) <- pcs$V2
22 | pcs <- pcs[,3:ncol(pcs)]
23 |
24 | # Make data frame of variables for model
25 | E <- as.numeric(exprs_raw[gene,])
26 | G <- t(subset(geno, ID == snp)[,as.character(meta$donor)])[,1]
27 | IND <- factor(meta$donor)
28 | B <- meta$batch
29 | AGE <- scale(meta$tbru_age)
30 | SEX <- meta$Sex
31 | nUMI <- scale(log(meta$nUMI))
32 | MT <- meta$percent_mito
33 | PC <- pcs[as.character(meta$donor),1:5]
34 | expPC <- pca_res[,1:5]
35 |
36 | data <- data.frame(E,G,IND,B,AGE,SEX,nUMI,MT,
37 | PC1 = PC[,1],PC2 = PC[,2],PC3 = PC[,3],PC4 = PC[,4],PC5 = PC[,5],
38 | expPC1 = expPC[,1], expPC2 = expPC[,2], expPC3 = expPC[,3], expPC4 = expPC[,4], expPC5 = expPC[,5])
39 | data$G <- as.numeric(as.character(data$G))
40 |
41 | full_model <- lme4::glmer(formula = E ~ G + (1 | IND) + (1 | B) + AGE + SEX + nUMI + MT + PC1 + PC2 + PC3 +PC4 + PC5 + expPC1 + expPC2 + expPC3 + expPC4 + expPC5,
42 | family = "poisson", nAGQ=0, data= data, control = glmerControl(optimizer = "nloptwrap"))
43 | null_model <- lme4::glmer(formula = E ~ (1 | IND) + (1 | B) + AGE + SEX + nUMI + MT + PC1 + PC2 + PC3 + PC4 + PC5 + expPC1 + expPC2 + expPC3 + expPC4 + expPC5,
44 | family = "poisson", nAGQ=0, data= data, control = glmerControl(optimizer = "nloptwrap"))
45 | model_lrt <- anova(null_model, full_model)
46 |
47 | out <- summary(test)$coefficients
48 | colnames(out) <- c("Estimate","Std.Error","zvalue","pval")
49 | out <- data.frame(gene=gene,snp=snp,term=row.names(x),x, lrt_pval=model_lrt$`Pr(>Chisq)`[2])
50 |
--------------------------------------------------------------------------------
/scripts/singlecell/poisson_nostate_permute.R:
--------------------------------------------------------------------------------
1 | # Script to run single-cell Poisson mixed effects eQTL model without cell state interaction on permuted data (permuting genotype)
2 |
3 | # Load libraries
4 | library(lme4)
5 | library(Matrix)
6 |
7 | args = commandArgs(trailingOnly=TRUE)
8 |
9 | gene <- args[1]
10 | snp <- args[2]
11 |
12 | # Load raw UMI counts matrix and metadata (from GEO: GSE158769) and gene expression PCs
13 | exprs_raw <- fread("/path/GSE158769_exprs_raw.tsv.gz")
14 | meta <- fread("/path/GSE158769_meta_data.txt.gz")
15 | pca_res <- readRDS("/path/pca_irlba.rds")
16 | pca_res <- pca_res$x
17 |
18 | # Load genotype dosages (from dbGaP: phs002025) and PCs
19 | geno <- fread("/path/geno_imputed_dosage_maf05_hg38_QC.vcf.gz")
20 | pcs <- read.delim("/path/geno_PCs.txt", sep = " ", header = F)
21 | row.names(pcs) <- pcs$V2
22 | pcs <- pcs[,3:ncol(pcs)]
23 |
24 | # Make data frame of variables for model
25 | set.seed(1010)
26 | E <- as.numeric(exprs_raw[gene,])
27 | G <- subset(geno, ID == snp)
28 | colnames(G)[10:276] <- sample(colnames(G)[10:276])
29 | G <- t(G[,as.character(meta$donor)])[,1]
30 | IND <- factor(meta$donor)
31 | B <- meta$batch
32 | AGE <- scale(meta$tbru_age)
33 | SEX <- meta$Sex
34 | nUMI <- scale(log(meta$nUMI))
35 | MT <- meta$percent_mito
36 | PC <- pcs[as.character(meta$donor),1:5]
37 | expPC <- pca_res[,1:5]
38 |
39 | data <- data.frame(E,G,IND,B,AGE,SEX,nUMI,MT,
40 | PC1 = PC[,1],PC2 = PC[,2],PC3 = PC[,3],PC4 = PC[,4],PC5 = PC[,5],
41 | expPC1 = expPC[,1], expPC2 = expPC[,2], expPC3 = expPC[,3], expPC4 = expPC[,4], expPC5 = expPC[,5])
42 | data$G <- as.numeric(as.character(data$G))
43 |
44 | full_model <- lme4::glmer(formula = E ~ G + (1 | IND) + (1 | B) + AGE + SEX + nUMI + MT + PC1 + PC2 + PC3 +PC4 + PC5 + expPC1 + expPC2 + expPC3 + expPC4 + expPC5,
45 | family = "poisson", nAGQ=0, data= data, control = glmerControl(optimizer = "nloptwrap"))
46 | null_model <- lme4::glmer(formula = E ~ (1 | IND) + (1 | B) + AGE + SEX + nUMI + MT + PC1 + PC2 + PC3 + PC4 + PC5 + expPC1 + expPC2 + expPC3 + expPC4 + expPC5,
47 | family = "poisson", nAGQ=0, data= data, control = glmerControl(optimizer = "nloptwrap"))
48 | model_lrt <- anova(null_model, full_model)
49 |
50 | out <- summary(test)$coefficients
51 | colnames(out) <- c("Estimate","Std.Error","zvalue","pval")
52 | out <- data.frame(gene=gene,snp=snp,term=row.names(x),x, lrt_pval=model_lrt$`Pr(>Chisq)`[2])
53 |
--------------------------------------------------------------------------------
/scripts/singlecell/poisson_univariate.R:
--------------------------------------------------------------------------------
1 | # Script to run single-cell Poisson mixed effects eQTL model with one cell state interaction (e.g., CV, PC, CD4+, trajectory, disease status)
2 |
3 | # Load libraries
4 | library(lme4)
5 | library(Matrix)
6 |
7 | args = commandArgs(trailingOnly=TRUE)
8 |
9 | gene <- args[1]
10 | snp <- args[2]
11 |
12 | # Load raw UMI counts matrix and metadata (from GEO: GSE158769) and gene expression PCs
13 | exprs_raw <- fread("/path/GSE158769_exprs_raw.tsv.gz")
14 | meta <- fread("/path/GSE158769_meta_data.txt.gz")
15 | pca_res <- readRDS("/path/pca_irlba.rds")
16 | pca_res <- pca_res$x
17 |
18 | # Load genotype dosages (from dbGaP: phs002025) and PCs
19 | geno <- fread("/path/geno_imputed_dosage_maf05_hg38_QC.vcf.gz")
20 | pcs <- read.delim("/path/geno_PCs.txt", sep = " ", header = F)
21 | row.names(pcs) <- pcs$V2
22 | pcs <- pcs[,3:ncol(pcs)]
23 |
24 | # Make data frame of variables for model
25 | E <- as.numeric(exprs_raw[gene,])
26 | G <- t(subset(geno, ID == snp)[,as.character(meta$donor)])[,1]
27 | IND <- factor(meta$donor)
28 | B <- meta$batch
29 | AGE <- scale(meta$tbru_age)
30 | SEX <- meta$Sex
31 | nUMI <- scale(log(meta$nUMI))
32 | MT <- meta$percent_mito
33 | PC <- pcs[as.character(meta$donor),1:5]
34 | expPC <- pca_res[,1:5]
35 |
36 | # Load interaction covariate (e.g., batch-corrected CV from CCA)
37 | cov <- readRDS("/path/cca_donor_batch.rds")[,1]
38 |
39 | data <- data.frame(E,G,IND,B,AGE,SEX,nUMI,MT,
40 | PC1 = PC[,1],PC2 = PC[,2],PC3 = PC[,3],PC4 = PC[,4],PC5 = PC[,5],
41 | expPC1 = expPC[,1], expPC2 = expPC[,2], expPC3 = expPC[,3], expPC4 = expPC[,4], expPC5 = expPC[,5], cov)
42 | data$G <- as.numeric(as.character(data$G))
43 |
44 | full_model <- lme4::glmer(formula = E ~ G + (1 | B) + (1 | IND) + AGE + SEX + nUMI + MT + PC1 + PC2 + PC3 + PC4 + PC5 + expPC1 + expPC2 + expPC3 + expPC4 + expPC5 + cov + G*cov,
45 | family = "poisson", nAGQ = 0, data= data, control = glmerControl(optimizer = "nloptwrap"))
46 | null_model <- lme4::glmer(formula = E ~ G + (1 | B) + (1 | IND) + AGE + SEX + nUMI + MT + PC1 + PC2 + PC3 + PC4 + PC5 + expPC1 + expPC2 + expPC3 + expPC4 + expPC5 + cov,
47 | family = "poisson", nAGQ = 0, data= data, control = glmerControl(optimizer = "nloptwrap"))
48 | model_lrt <- anova(null_model, full_model)
49 |
50 | out <- summary(test)$coefficients
51 | colnames(out) <- c("Estimate","Std.Error","zvalue","pval")
52 | out <- data.frame(gene=gene,snp=snp,term=row.names(x),x, lrt_pval=model_lrt$`Pr(>Chisq)`[2])
53 |
--------------------------------------------------------------------------------
/scripts/singlecell/poisson_univariate_condition.R:
--------------------------------------------------------------------------------
1 | # Script to run single-cell Poisson mixed effects eQTL conditional model with CV1 and CD4+ cell states
2 |
3 | # Load libraries
4 | library(lme4)
5 | library(Matrix)
6 |
7 | args = commandArgs(trailingOnly=TRUE)
8 |
9 | gene <- args[1]
10 | snp <- args[2]
11 |
12 | # Load raw UMI counts matrix and metadata (from GEO: GSE158769) and gene expression PCs
13 | exprs_raw <- fread("/path/GSE158769_exprs_raw.tsv.gz")
14 | meta <- fread("/path/GSE158769_meta_data.txt.gz")
15 | pca_res <- readRDS("/path/pca_irlba.rds")
16 | pca_res <- pca_res$x
17 |
18 | # Load genotype dosages (from dbGaP: phs002025) and PCs
19 | geno <- fread("/path/geno_imputed_dosage_maf05_hg38_QC.vcf.gz")
20 | pcs <- read.delim("/path/geno_PCs.txt", sep = " ", header = F)
21 | row.names(pcs) <- pcs$V2
22 | pcs <- pcs[,3:ncol(pcs)]
23 |
24 | # Make data frame of variables for model
25 | E <- as.numeric(exprs_raw[gene,])
26 | G <- t(subset(geno, ID == snp)[,as.character(meta$donor)])[,1]
27 | IND <- factor(meta$donor)
28 | B <- meta$batch
29 | AGE <- scale(meta$tbru_age)
30 | SEX <- meta$Sex
31 | nUMI <- scale(log(meta$nUMI))
32 | MT <- meta$percent_mito
33 | PC <- pcs[as.character(meta$donor),1:5]
34 | expPC <- pca_res[,1:5]
35 |
36 | # Load interaction covariates (batch-corrected CV from CCA and CD4+ binary annotation from gating on CITE-seq surface protein)
37 | cov <- readRDS("/path/cca_donor_batch.rds")[,1]
38 | cd4 <- readRDS("/path/citeseq_gates.rds")[,"cd4"]
39 |
40 | data <- data.frame(E,G,IND,B,AGE,SEX,nUMI,MT,
41 | PC1 = PC[,1],PC2 = PC[,2],PC3 = PC[,3],PC4 = PC[,4],PC5 = PC[,5],
42 | expPC1 = expPC[,1], expPC2 = expPC[,2], expPC3 = expPC[,3], expPC4 = expPC[,4], expPC5 = expPC[,5], cov, cd4)
43 | data$G <- as.numeric(as.character(data$G))
44 |
45 | full_model <- lme4::glmer(formula = E ~ G + (1 | B) + (1 | IND) + AGE + SEX + nUMI + MT + PC1 + PC2 + PC3 + PC4 + PC5 + expPC1 + expPC2 + expPC3 + expPC4 + expPC5 + cov + G*cov + cd4 + G*cd4,
46 | family = "poisson", nAGQ = 0, data= data, control = glmerControl(optimizer = "nloptwrap"))
47 | null_model_condCD4 <- lme4::glmer(formula = E ~ G + (1 | B) + (1 | IND) + AGE + SEX + nUMI + MT + PC1 + PC2 + PC3 + PC4 + PC5 + expPC1 + expPC2 + expPC3 + expPC4 + expPC5 + cov + cd4 + G*cd4,
48 | family = "poisson", nAGQ = 0, data= data, control = glmerControl(optimizer = "nloptwrap"))
49 | model_lrt_condCD4 <- anova(null_model_condCD4, full_model)
50 | null_model_condCV1 <- lme4::glmer(formula = E ~ G + (1 | B) + (1 | IND) + AGE + SEX + nUMI + MT + PC1 + PC2 + PC3 + PC4 + PC5 + expPC1 + expPC2 + expPC3 + expPC4 + expPC5 + cov + G*cov + cd4,
51 | family = "poisson", nAGQ = 0, data= data, control = glmerControl(optimizer = "nloptwrap"))
52 | model_lrt_condCV1 <- anova(null_model_condCV1, full_model)
53 |
--------------------------------------------------------------------------------
/scripts/singlecell/poisson_univariate_permute.R:
--------------------------------------------------------------------------------
1 | # Script to run single-cell Poisson mixed effects eQTL model with one cell state interaction (e.g., CV, PC, CD4+, trajectory, disease status) on permuted data (permuting cell state)
2 |
3 | # Load libraries
4 | library(lme4)
5 | library(Matrix)
6 |
7 | args = commandArgs(trailingOnly=TRUE)
8 |
9 | gene <- args[1]
10 | snp <- args[2]
11 |
12 | # Load raw UMI counts matrix and metadata (from GEO: GSE158769) and gene expression PCs
13 | exprs_raw <- fread("/path/GSE158769_exprs_raw.tsv.gz")
14 | meta <- fread("/path/GSE158769_meta_data.txt.gz")
15 | pca_res <- readRDS("/path/pca_irlba.rds")
16 | pca_res <- pca_res$x
17 |
18 | # Load genotype dosages (from dbGaP: phs002025) and PCs
19 | geno <- fread("/path/geno_imputed_dosage_maf05_hg38_QC.vcf.gz")
20 | pcs <- read.delim("/path/geno_PCs.txt", sep = " ", header = F)
21 | row.names(pcs) <- pcs$V2
22 | pcs <- pcs[,3:ncol(pcs)]
23 |
24 | # Load interaction covariate (e.g., batch-corrected CV from CCA)
25 | set.seed(1010)
26 | cov <- unlist(readRDS("/path/cca_donor_batch.rds")[,1])
27 | for(x in unique(meta$donor)){
28 | cov[meta$donor == x] <- sample(cov[meta$donor == x])
29 | }
30 |
31 | # Make data frame of variables for model
32 | E <- as.numeric(exprs_raw[gene,])
33 | G <- t(subset(geno, ID == snp)[,as.character(meta$donor)])[,1]
34 | IND <- factor(meta$donor)
35 | B <- meta$batch
36 | AGE <- scale(meta$tbru_age)
37 | SEX <- meta$Sex
38 | nUMI <- scale(log(meta$nUMI))
39 | MT <- meta$percent_mito
40 | PC <- pcs[as.character(meta$donor),1:5]
41 | expPC <- pca_res[,1:5]
42 |
43 | data <- data.frame(E,G,IND,B,AGE,SEX,nUMI,MT,
44 | PC1 = PC[,1],PC2 = PC[,2],PC3 = PC[,3],PC4 = PC[,4],PC5 = PC[,5],
45 | expPC1 = expPC[,1], expPC2 = expPC[,2], expPC3 = expPC[,3], expPC4 = expPC[,4], expPC5 = expPC[,5], cov)
46 | data$G <- as.numeric(as.character(data$G))
47 |
48 | full_model <- lme4::glmer(formula = E ~ G + (1 | B) + (1 | IND) + AGE + SEX + nUMI + MT + PC1 + PC2 + PC3 + PC4 + PC5 + expPC1 + expPC2 + expPC3 + expPC4 + expPC5 + cov + G*cov,
49 | family = "poisson", nAGQ = 0, data= data, control = glmerControl(optimizer = "nloptwrap"))
50 | null_model <- lme4::glmer(formula = E ~ G + (1 | B) + (1 | IND) + AGE + SEX + nUMI + MT + PC1 + PC2 + PC3 + PC4 + PC5 + expPC1 + expPC2 + expPC3 + expPC4 + expPC5 + cov,
51 | family = "poisson", nAGQ = 0, data= data, control = glmerControl(optimizer = "nloptwrap"))
52 | model_lrt <- anova(null_model, full_model)
53 |
54 | out <- summary(test)$coefficients
55 | colnames(out) <- c("Estimate","Std.Error","zvalue","pval")
56 | out <- data.frame(gene=gene,snp=snp,term=row.names(x),x, lrt_pval=model_lrt$`Pr(>Chisq)`[2])
57 |
--------------------------------------------------------------------------------
/scripts/singlecell/poisson_univariate_simulDE.R:
--------------------------------------------------------------------------------
1 | # Script to run single-cell Poisson mixed effects eQTL model with one cell state interaction (CD4+), simulating differential expression
2 |
3 | # Load libraries
4 | library(lme4)
5 | library(Matrix)
6 |
7 | args = commandArgs(trailingOnly=TRUE)
8 |
9 | gene <- args[1]
10 | snp <- args[2]
11 | frac <- as.numeric(args[3]) # fraction to which to decrease expression in CD4+ cells
12 |
13 | # Load raw UMI counts matrix and metadata (from GEO: GSE158769) and gene expression PCs
14 | exprs_raw <- fread("/path/GSE158769_exprs_raw.tsv.gz")
15 | meta <- fread("/path/GSE158769_meta_data.txt.gz")
16 | pca_res <- readRDS("/path/pca_irlba.rds")
17 | pca_res <- pca_res$x
18 |
19 | # Load genotype dosages (from dbGaP: phs002025) and PCs
20 | geno <- fread("/path/geno_imputed_dosage_maf05_hg38_QC.vcf.gz")
21 | pcs <- read.delim("/path/geno_PCs.txt", sep = " ", header = F)
22 | row.names(pcs) <- pcs$V2
23 | pcs <- pcs[,3:ncol(pcs)]
24 |
25 | # Make data frame of variables for model
26 | E <- as.numeric(exprs_raw[gene,])
27 | G <- t(subset(geno, ID == snp)[,as.character(meta$donor)])[,1]
28 | IND <- factor(meta$donor)
29 | B <- meta$batch
30 | AGE <- scale(meta$tbru_age)
31 | SEX <- meta$Sex
32 | nUMI <- scale(log(meta$nUMI))
33 | MT <- meta$percent_mito
34 | PC <- pcs[as.character(meta$donor),1:5]
35 | expPC <- pca_res[,1:5]
36 |
37 | # Load interaction covariate (e.g., batch-corrected CV from CCA)
38 | cov <- readRDS("/path/citeseq_gates.rds")[,"cd4"]
39 |
40 | data <- data.frame(E,G,IND,B,AGE,SEX,nUMI,MT,
41 | PC1 = PC[,1],PC2 = PC[,2],PC3 = PC[,3],PC4 = PC[,4],PC5 = PC[,5],
42 | expPC1 = expPC[,1], expPC2 = expPC[,2], expPC3 = expPC[,3], expPC4 = expPC[,4], expPC5 = expPC[,5], cov)
43 | data$G <- as.numeric(as.character(data$G))
44 | data$E[data$cov == TRUE] <- round(2^(log2(data$E[data$cov == TRUE]+1)/frac)-1)
45 |
46 | full_model <- lme4::glmer(formula = E ~ G + (1 | B) + (1 | IND) + AGE + SEX + nUMI + MT + PC1 + PC2 + PC3 + PC4 + PC5 + expPC1 + expPC2 + expPC3 + expPC4 + expPC5 + cov + G*cov,
47 | family = "poisson", nAGQ = 0, data= data, control = glmerControl(optimizer = "nloptwrap"))
48 | null_model <- lme4::glmer(formula = E ~ G + (1 | B) + (1 | IND) + AGE + SEX + nUMI + MT + PC1 + PC2 + PC3 + PC4 + PC5 + expPC1 + expPC2 + expPC3 + expPC4 + expPC5 + cov,
49 | family = "poisson", nAGQ = 0, data= data, control = glmerControl(optimizer = "nloptwrap"))
50 | model_lrt <- anova(null_model, full_model)
51 |
52 | out <- summary(test)$coefficients
53 | colnames(out) <- c("Estimate","Std.Error","zvalue","pval")
54 | out <- data.frame(gene=gene,snp=snp,term=row.names(x),x, lrt_pval=model_lrt$`Pr(>Chisq)`[2])
55 |
--------------------------------------------------------------------------------
/scripts/singlecell/poisson_univariate_simulnull.R:
--------------------------------------------------------------------------------
1 | # Script to run single-cell Poisson mixed effects eQTL model with one cell state interaction (e.g., CV, PC, CD4+, trajectory, disease status) and simulated null (based on Buzkova, et al. 2011)
2 |
3 | # Load libraries
4 | library(lme4)
5 | library(Matrix)
6 |
7 | args = commandArgs(trailingOnly=TRUE)
8 |
9 | gene <- args[1]
10 | snp <- args[2]
11 |
12 | # Load raw UMI counts matrix and metadata (from GEO: GSE158769) and gene expression PCs
13 | exprs_raw <- fread("/path/GSE158769_exprs_raw.tsv.gz")
14 | meta <- fread("/path/GSE158769_meta_data.txt.gz")
15 | pca_res <- readRDS("/path/pca_irlba.rds")
16 | pca_res <- pca_res$x
17 |
18 | # Load genotype dosages (from dbGaP: phs002025) and PCs
19 | geno <- fread("/path/geno_imputed_dosage_maf05_hg38_QC.vcf.gz")
20 | pcs <- read.delim("/path/geno_PCs.txt", sep = " ", header = F)
21 | row.names(pcs) <- pcs$V2
22 | pcs <- pcs[,3:ncol(pcs)]
23 |
24 | # Make data frame of variables for model
25 | E <- as.numeric(exprs_raw[gene,])
26 | G <- t(subset(geno, ID == snp)[,as.character(meta$donor)])[,1]
27 | IND <- factor(meta$donor)
28 | B <- meta$batch
29 | AGE <- scale(meta$tbru_age)
30 | SEX <- meta$Sex
31 | nUMI <- scale(log(meta$nUMI))
32 | MT <- meta$percent_mito
33 | PC <- pcs[as.character(meta$donor),1:5]
34 | expPC <- pca_res[,1:5]
35 |
36 | # Load interaction covariate (e.g., batch-corrected CV from CCA)
37 | cov <- readRDS("/path/cca_donor_batch.rds")[,1]
38 |
39 | data <- data.frame(E,G,IND,B,AGE,SEX,nUMI,MT,
40 | PC1 = PC[,1],PC2 = PC[,2],PC3 = PC[,3],PC4 = PC[,4],PC5 = PC[,5],
41 | expPC1 = expPC[,1], expPC2 = expPC[,2], expPC3 = expPC[,3], expPC4 = expPC[,4], expPC5 = expPC[,5], cov)
42 | data$G <- as.numeric(as.character(data$G))
43 |
44 | null_model <- lme4::glmer(formula = E ~ G + (1 | B) + (1 | IND) + AGE + SEX + nUMI + MT + PC1 + PC2 + PC3 + PC4 + PC5 + expPC1 + expPC2 + expPC3 + expPC4 + expPC5 + cov,
45 | family = "poisson", nAGQ = 0, data= data, control = glmerControl(optimizer = "nloptwrap"))
46 | null_pred_y <- predict(null_model, data,type="response")
47 |
48 | set.seed(1010)
49 | null_betas <- sapply(1:1000, function(x) {
50 | y_vec <- sapply(1:500089, function(x){rpois(1, null_pred_y[x])})
51 | data$y <- y_vec
52 |
53 | full_model <- lme4::glmer(formula = y ~ G + (1 | IND) + (1 | B) + AGE + SEX + nUMI + MT + PC1 + PC2 + PC3 + PC4 + PC5 + expPC1 + expPC2 + expPC3 + expPC4 + expPC5 + cov + G*cov,
54 | family = "poisson", nAGQ=0, data= data, control = glmerControl(optimizer = "nloptwrap"))
55 |
56 | return(fixef(full_model)[18])
57 | })
58 |
59 | full_model <- lme4::glmer(formula = E ~ G + (1 | B) + (1 | IND) + AGE + SEX + nUMI + MT + PC1 + PC2 + PC3 + PC4 + PC5 + expPC1 + expPC2 + expPC3 + expPC4 + expPC5 + cov + G*cov,
60 | family = "poisson", nAGQ = 0, data= data, control = glmerControl(optimizer = "nloptwrap"))
61 |
62 | sum(abs(fixef(full_model)[18]) > abs(null_betas))
63 |
--------------------------------------------------------------------------------