├── .gitignore
├── README.md
├── prepare_ligands.py
├── prepare_receptor.py
├── run_autodock_vina.py
├── test_files
    └── ZINC67842136.mol2
└── tests.py


/.gitignore:
--------------------------------------------------------------------------------
 1 | # Byte-compiled / optimized / DLL files
 2 | __pycache__/
 3 | *.py[cod]
 4 | 
 5 | # C extensions
 6 | *.so
 7 | 
 8 | # Distribution / packaging
 9 | .Python
10 | env/
11 | bin/
12 | build/
13 | develop-eggs/
14 | dist/
15 | eggs/
16 | lib/
17 | lib64/
18 | parts/
19 | sdist/
20 | var/
21 | *.egg-info/
22 | .installed.cfg
23 | *.egg
24 | 
25 | # Installer logs
26 | pip-log.txt
27 | pip-delete-this-directory.txt
28 | 
29 | # Unit test / coverage reports
30 | htmlcov/
31 | .tox/
32 | .coverage
33 | .cache
34 | nosetests.xml
35 | coverage.xml
36 | 
37 | # Translations
38 | *.mo
39 | 
40 | # Mr Developer
41 | .mr.developer.cfg
42 | .project
43 | .pydevproject
44 | 
45 | # Rope
46 | .ropeproject
47 | 
48 | # Django stuff:
49 | *.log
50 | *.pot
51 | 
52 | # Sphinx documentation
53 | docs/_build/
54 | 
55 | # DockingWatch
56 | .DockingWatch.conf
57 | 


--------------------------------------------------------------------------------
/README.md:
--------------------------------------------------------------------------------
1 | docking-scripts
2 | ===============
3 | 
4 | Scripts used for virtual screening of small molecule inhibitors from ZINC database using Autodock Vina
5 | 


--------------------------------------------------------------------------------
/prepare_ligands.py:
--------------------------------------------------------------------------------
 1 | #!/usr/bin/env python
 2 | 
 3 | ##########################################################################
 4 | # Author: joemartaganna (joemar.ct@gmail.com)
 5 | #
 6 | # Description:
 7 | #
 8 | # Script for converting mol2 files downloaded from the ZINC database into 
 9 | # pdbqt files with charges and polar hydrogens added in preparation for 
10 | # docking with AutoDock Vina. The conversion is done in parallel through
11 | # multithreading.
12 | #
13 | # Requires: babel, joblib
14 | ##########################################################################
15 | 
16 | 
17 | from subprocess import Popen, PIPE
18 | from joblib import Parallel, delayed
19 | import gzip
20 | import os
21 | 
22 |     
23 | def convert_to_pdbqt(mol2, output_dir, remove_input=True):
24 |     mol_id = mol2.split('\n')[1]
25 |     # Write the mol2
26 |     mol2_outf_path = os.path.join(output_dir, mol_id + '.mol2')
27 |     mol2_outf = open(mol2_outf_path, 'w')
28 |     mol2_outf.write(mol2)
29 |     mol2_outf.close()
30 |     # Convert to pdbqt
31 |     pdbqt_outf_path = mol2_outf_path.replace('.mol2', '.pdbqt')
32 |     cmd = 'babel -imol2 %s -opdbqt %s --partialcharge gasteiger --AddPolarH' % (mol2_outf_path, pdbqt_outf_path)
33 |     p = Popen(cmd, shell=True, stdout=PIPE, stderr=PIPE)
34 |     stdout, stderr = p.communicate()
35 |     if remove_input:
36 |         os.remove(mol2_outf_path)
37 |     return {'mol2': mol2_outf_path, 'pdbqt': pdbqt_outf_path}
38 |             
39 | 
40 | def split_gzipped_mol2(gzipped_mol2, output_dir=None):
41 |     f = gzip.open(gzipped_mol2, 'rb').read()
42 |     p = '@<TRIPOS>MOLECULE'
43 |     mols_count = f.count(p)
44 |     mols = f.split(p)
45 |     if not output_dir:
46 |         output_dir = gzipped_mol2.split('.mol2.gz')[0]
47 |         if not os.path.exists(output_dir):
48 |             os.mkdir(output_dir)
49 |     def format_mol(mol):
50 |         mol = [p] + [x.strip() for x in mol.split('\n') if x]
51 |         mol = '\n'.join(mol)
52 |         return mol
53 |     mols = [format_mol(x) for x in mols if x]
54 |     return mols
55 |     
56 | 
57 | def generate_pdbqt_files(gzipped_mol2, mols):
58 |     output_dir = gzipped_mol2.split('.mol2.gz')[0]
59 |     Parallel(n_jobs=50, backend='threading', verbose=55)(delayed(convert_to_pdbqt)(mol, output_dir) for mol in mols)
60 |     
61 | 
62 | if __name__ == '__main__':
63 |     import sys
64 |     infiles = sys.argv[1:]
65 |     for i, f in enumerate(infiles):
66 |         print '\n\n%s/%s - dealing with %s' % (i+1, len(infiles), f)
67 |         mols = split_gzipped_mol2(f)
68 |         generate_pdbqt_files(f, mols)


--------------------------------------------------------------------------------
/prepare_receptor.py:
--------------------------------------------------------------------------------
 1 | #!/usr/bin/env python
 2 | 
 3 | import os
 4 | import sys
 5 | import glob
 6 | from subprocess import Popen, PIPE
 7 | 
 8 | 
 9 | def check_mgl_library():
10 |     if os.path.exists('/Library/MGLTools'):
11 |         MGL_LIB = glob.glob('/Library/MGLTools/*')
12 |         pythonpath = os.path.join(MGL_LIB[0], 'bin', 'pythonsh')
13 |         utilities24 = os.path.join(MGL_LIB[0], 'MGLToolsPckgs', 'AutoDockTools', 'Utilities24')
14 |         prep_receptor_script = os.path.join(utilities24, 'prepare_receptor4.py')
15 |         return (pythonpath, prep_receptor_script)
16 |     else:
17 |         raise Exception('The prepare_receptor4.py script cannot be found because MGLTools is not installed.')
18 | 
19 | 
20 | def execute_receptor_prep(receptor_pdb, output_pdbqt):
21 |     pythonpath, prep_script = check_mgl_library()
22 |     args = (pythonpath, prep_script, receptor_pdb, output_pdbqt)
23 |     cmd = '%s %s -r %s -o %s -A checkhydrogens' % args
24 |     p = Popen(cmd, shell=True, stdout=PIPE)
25 |     stdout, stderr = p.communicate()
26 |     print 'Done!'
27 | 
28 | 
29 | if __name__ == '__main__':
30 |     try:
31 |         receptor_pdb, output_pdbqt = sys.argv[1:]
32 |         execute_receptor_prep(receptor_pdb, output_pdbqt)
33 |     except ValueError:
34 |         print '\nUsage: prepare_receptor.py <receptor_pdb> <output_pdbqt>\n'


--------------------------------------------------------------------------------
/run_autodock_vina.py:
--------------------------------------------------------------------------------
  1 | #!/usr/bin/env python
  2 | 
  3 | ##########################################################################
  4 | # Author: joemartaganna (joemar.ct@gmail.com)
  5 | #
  6 | # Description:
  7 | #
  8 | # Script for running AutoDock Vina directly against gzipped mol2 files
  9 | # to be more memory- and disk space-efficient. Individual mol2 files are
 10 | # extracted from a gzipped file downloaded from the ZINC database. Each
 11 | # file is then converted to pdbqt, then docked against a receptor protein.
 12 | # An affinity cut-off is set so that the intermediate files produced
 13 | # during docking with results below this cut-off are discarded. Only the
 14 | # output files from high affinity conformers (very few) are stored. 
 15 | #
 16 | # The ZINC id of the ligands docked are saved into a Redis database. Each
 17 | # docking run checks this database to avoid redocking the same ligands 
 18 | # if the script is rerun because of errors or accidental shutdown.
 19 | #
 20 | # There is an optional support for delivery of binders into Dropbox. This
 21 | # is useful for easy monitoring of results in cases when the virtual
 22 | # screening is run remotely on a server.
 23 | #
 24 | # Requires: vina, babel, joblib, redis, redispy, prepare_ligands
 25 | # Optional: dropbox
 26 | ##########################################################################
 27 | 
 28 | import os
 29 | import sys
 30 | import json
 31 | import redis
 32 | import shutil
 33 | from joblib import Parallel, delayed
 34 | from subprocess import Popen, PIPE
 35 | from prepare_ligands import split_gzipped_mol2, convert_to_pdbqt
 36 | 
 37 | 
 38 | try:
 39 |     import dropbox
 40 |     dropbox_installed = True
 41 | except ImportError:
 42 |     dropbox_installed = False
 43 |     
 44 |     
 45 | def setup_dropbox_integration():
 46 |     script_path = os.path.realpath(__file__)
 47 |     conf_path = os.path.join(os.path.split(script_path)[0], '.DockingWatch.conf')
 48 |     app_key = '0raajjtnfqaq6ag'
 49 |     app_secret = 'ruc4zlu5gyk0rn3'
 50 |     flow = dropbox.client.DropboxOAuth2FlowNoRedirect(app_key, app_secret)
 51 |     if os.path.exists(conf_path):
 52 |         dbox_args = json.load(open(conf_path, 'r'))
 53 |         access_token = dbox_args['access_token']
 54 |         folder = dbox_args['folder']
 55 |     else:
 56 |         authorize_url = flow.start()
 57 |         # Authorize the Dropbox integration
 58 |         msg = '\nFor Dropbox integration, visit the following URL in your browser, '
 59 |         msg += 'grant access, and copy the resulting authorization code.'
 60 |         msg += ' Just hit enter without the code to skip.'
 61 |         print msg, '\n'
 62 |         print authorize_url
 63 |         code = raw_input("\nEnter the authorization code here: ").strip()
 64 |         folder = raw_input("Enter custom folder under DockingWatch app: ").strip()
 65 |         access_token, user_id = flow.finish(code)
 66 |         conf = {'access_token': access_token, 'folder': folder}
 67 |         json.dump(conf, open(conf_path, 'w'))
 68 |     return access_token, folder
 69 |     
 70 |     
 71 | def upload_to_dropbox(f, client, folder):
 72 |     fname = os.path.split(f)[-1]
 73 |     f = open(f, 'r')
 74 |     client.put_file(folder+'/binders/'+fname, f)
 75 |     
 76 | 
 77 | def dock_ligand(vina_conf, mol2_string, output_dir, affinity_cutoff):
 78 |     global db
 79 |     global dropbox_installed
 80 |     mol_id = mol2_string.split('\n')[1]
 81 |     try:
 82 |         # Check if this has been docked
 83 |         db[mol_id]
 84 |     except KeyError:        
 85 |         temp_dir = os.path.join(output_dir, 'temp')
 86 |         if not os.path.exists(temp_dir):
 87 |             os.mkdir(temp_dir)
 88 |         ligand_paths = convert_to_pdbqt(mol2_string, temp_dir)
 89 |         ligand = ligand_paths['pdbqt']
 90 |         cmd = 'vina --config %s --ligand %s' % (vina_conf, ligand)
 91 |         p = Popen(cmd, shell=True, stdout=PIPE, stderr=PIPE)
 92 |         stdout, stderr = p.communicate()
 93 |         if 'Refining results ... done' in stdout:
 94 |             lines = stdout.splitlines()
 95 |             best = None
 96 |             for i, line in enumerate(lines):
 97 |                 if '-+-' in line:
 98 |                     best = lines[i+1]
 99 |             if best:
100 |                 affinity = float(best.split()[1]) * -1
101 |             else:
102 |                 affinity = 0
103 |             # Put this into the record
104 |             db[mol_id] = affinity
105 |             # Remove the input and output files if affinity is below cut-off
106 |             docking_out = ligand.replace('.pdbqt', '_out.pdbqt')
107 |             if affinity < affinity_cutoff:
108 |                 os.remove(ligand)
109 |                 os.remove(docking_out)
110 |             else:
111 |                 binders_dir = os.path.join(output_dir, 'binders')
112 |                 if not os.path.exists(binders_dir):
113 |                     os.mkdir(binders_dir)
114 |                 shutil.move(ligand, binders_dir)
115 |                 shutil.move(ligand.replace('.pdbqt', '_out.pdbqt'), binders_dir)
116 |                 if dropbox_installed:
117 |                     global client
118 |                     global dropbox_folder
119 |                     upload_to_dropbox(os.path.join(binders_dir, mol_id+'.pdbqt'), client, dropbox_folder)
120 |                     upload_to_dropbox(os.path.join(binders_dir, mol_id+'_out.pdbqt'), client, dropbox_folder)
121 |         else:
122 |             print 'Error on %s: The conversion from mol2 to pdbqt may not have succeeded.' % mol_id
123 |         
124 |         
125 | def execute_virtual_screening(vina_conf, gzipped_mol2, output_dir=None, affinity_cutoff=0):
126 |     mols = split_gzipped_mol2(gzipped_mol2, output_dir=output_dir)
127 |     print 'There are %s ligands in this file.' % len(mols)
128 |     Parallel(n_jobs=-1, backend='multiprocessing', verbose=55)(delayed(dock_ligand)(vina_conf, mol, output_dir, affinity_cutoff) for mol in mols)
129 |     
130 |     
131 | def connect_to_redisdb():
132 |     db = redis.Redis(host='localhost', port=6379, db=7)
133 |     return db
134 | 
135 |         
136 | if __name__ == '__main__':
137 | 
138 |     conf = sys.argv[1]
139 |     infiles = sys.argv[2:]
140 |     # Set the output directory
141 |     output_dir = os.path.join(os.getcwd(), 'docking_results')
142 |     if not os.path.exists(output_dir):
143 |         os.mkdir(output_dir)
144 |     # Start the redis db
145 |     redis_conf = os.path.join(os.getcwd(), 'docking_results', 'redis.conf')
146 |     # Setup Dropbox integration
147 |     if dropbox_installed:
148 |         access_token, dropbox_folder = setup_dropbox_integration()
149 |         if access_token:
150 |             client = dropbox.client.DropboxClient(access_token)
151 |     # Execute virtual screening
152 |     if os.path.exists(redis_conf):
153 |         db = connect_to_redisdb()
154 |         for i, f in enumerate(infiles):
155 |             print '\n\n%s/%s - dealing with %s' % (i+1, len(infiles), f)
156 |             execute_virtual_screening(conf, f, output_dir=output_dir, affinity_cutoff=8.0)
157 |     else:
158 |         rc = open(redis_conf, 'w')
159 |         rc.write('dbfilename records.rdb\ndir %s\nsave 900 1' % output_dir)
160 |         rc.close()
161 |         print '\nRedis configuration file has been created.'
162 |         print 'Run redis-server using the redis.conf file and rerun this script to start the virtual screening.\n'
163 |             
164 | 
165 | 


--------------------------------------------------------------------------------
/test_files/ZINC67842136.mol2:
--------------------------------------------------------------------------------
  1 | @<TRIPOS>MOLECULE
  2 | ZINC67842136
  3 | 49    52     0     0     0
  4 | SMALL
  5 | USER_CHARGES
  6 | @<TRIPOS>ATOM
  7 | 1 C1          0.2282    2.4095   -4.5870 C.3       1 <0>        -0.1014
  8 | 2 C2         -0.6279    2.2038   -3.3639 C.ar      1 <0>        -0.0179
  9 | 3 C3         -1.9904    2.4551   -3.4298 C.ar      1 <0>         0.2711
 10 | 4 C4         -2.7691    2.2606   -2.2932 C.ar      1 <0>        -0.1151
 11 | 5 C5         -2.1467    1.8212   -1.1361 C.ar      1 <0>        -0.0399
 12 | 6 C6         -0.7833    1.5922   -1.1482 C.ar      1 <0>        -0.0060
 13 | 7 N1         -0.0750    1.7816   -2.2440 N.ar      1 <0>        -0.3881
 14 | 8 C7         -2.9464    1.5960    0.1212 C.3       1 <0>         0.1212
 15 | 9 O1         -2.0632    1.2639    1.1944 O.3       1 <0>        -0.5758
 16 | 10 C8         -4.2530    2.5219   -2.3200 C.3       1 <0>         0.1049
 17 | 11 N2         -4.5223    3.8696   -1.8013 N.3       1 <0>        -0.5303
 18 | 12 C9         -4.0885    4.8974   -2.7562 C.3       1 <0>         0.0579
 19 | 13 C10        -4.2493    6.2823   -2.1251 C.3       1 <0>        -0.1147
 20 | 14 C11        -5.7172    6.4979   -1.7464 C.3       1 <0>        -0.1115
 21 | 15 C12        -6.1577    5.3868   -0.7894 C.3       1 <0>         0.0099
 22 | 16 H1         -5.5677    5.4395    0.1256 H         1 <0>         0.1093
 23 | 17 C13        -5.9416    4.0279   -1.4592 C.3       1 <0>         0.0655
 24 | 18 C14        -7.6176    5.5584   -0.4573 C.2       1 <0>         0.2569
 25 | 19 N3         -8.5839    5.5383   -1.3253 N.2       1 <0>        -0.4717
 26 | 20 C15        -9.7634    5.7233   -0.6728 C.ar      1 <0>         0.0214
 27 | 21 C16       -11.0952    5.7916   -1.0851 C.ar      1 <0>        -0.0613
 28 | 22 C17       -12.0887    5.9942   -0.1536 C.ar      1 <0>        -0.1193
 29 | 23 C18       -11.7741    6.1311    1.1904 C.ar      1 <0>        -0.0930
 30 | 24 C19       -10.4626    6.0663    1.6120 C.ar      1 <0>        -0.0931
 31 | 25 C20        -9.4493    5.8619    0.6876 C.ar      1 <0>         0.0061
 32 | 26 O2         -8.1061    5.7482    0.7786 O.3       1 <0>        -0.2127
 33 | 27 O3         -2.5568    2.8857   -4.5878 O.3       1 <0>        -0.6252
 34 | 28 H2          0.6009    3.4337   -4.6015 H         1 <0>         0.0820
 35 | 29 H3          1.0697    1.7170   -4.5621 H         1 <0>         0.0661
 36 | 30 H4         -0.3666    2.2265   -5.4818 H         1 <0>         0.1018
 37 | 31 H5         -0.2926    1.2499   -0.2491 H         1 <0>         0.1851
 38 | 32 H6         -3.4953    2.5041    0.3703 H         1 <0>         0.0832
 39 | 33 H7         -3.6495    0.7783   -0.0369 H         1 <0>         0.0670
 40 | 34 H8         -2.5103    1.1050    2.0370 H         1 <0>         0.3903
 41 | 35 H9         -4.6165    2.4477   -3.3450 H         1 <0>         0.0673
 42 | 36 H10        -4.7630    1.7851   -1.6994 H         1 <0>         0.0789
 43 | 37 H11        -3.0419    4.7351   -3.0140 H         1 <0>         0.0845
 44 | 38 H12        -4.6981    4.8361   -3.6577 H         1 <0>         0.0435
 45 | 39 H13        -3.6292    6.3491   -1.2311 H         1 <0>         0.0717
 46 | 40 H14        -3.9419    7.0458   -2.8397 H         1 <0>         0.0753
 47 | 41 H15        -5.8288    7.4656   -1.2572 H         1 <0>         0.0696
 48 | 42 H16        -6.3331    6.4700   -2.6453 H         1 <0>         0.0773
 49 | 43 H17        -6.5425    3.9707   -2.3668 H         1 <0>         0.0541
 50 | 44 H18        -6.2410    3.2340   -0.7749 H         1 <0>         0.0778
 51 | 45 H19       -11.3453    5.6863   -2.1304 H         1 <0>         0.1364
 52 | 46 H20       -13.1195    6.0471   -0.4715 H         1 <0>         0.1385
 53 | 47 H21       -12.5619    6.2900    1.9120 H         1 <0>         0.1415
 54 | 48 H22       -10.2252    6.1745    2.6600 H         1 <0>         0.1356
 55 | 49 H23         0.8794    1.6091   -2.2247 H         1 <0>         0.4251
 56 | @<TRIPOS>BOND
 57 | 1    1   28 1
 58 | 2    1   29 1
 59 | 3    1   30 1
 60 | 4    1    2 1
 61 | 5    2    3 ar
 62 | 6    2    7 ar
 63 | 7    3   27 1
 64 | 8    3    4 ar
 65 | 9    4    5 ar
 66 | 10    4   10 1
 67 | 11    5    6 ar
 68 | 12    5    8 1
 69 | 13    6    7 ar
 70 | 14    6   31 1
 71 | 15    7   49 1
 72 | 16    8    9 1
 73 | 17    8   32 1
 74 | 18    8   33 1
 75 | 19    9   34 1
 76 | 20   10   11 1
 77 | 21   10   35 1
 78 | 22   10   36 1
 79 | 23   11   17 1
 80 | 24   11   12 1
 81 | 25   12   13 1
 82 | 26   12   37 1
 83 | 27   12   38 1
 84 | 28   13   14 1
 85 | 29   13   39 1
 86 | 30   13   40 1
 87 | 31   14   15 1
 88 | 32   14   41 1
 89 | 33   14   42 1
 90 | 34   15   16 1
 91 | 35   15   17 1
 92 | 36   15   18 1
 93 | 37   17   43 1
 94 | 38   17   44 1
 95 | 39   18   26 1
 96 | 40   18   19 2
 97 | 41   19   20 1
 98 | 42   20   25 ar
 99 | 43   20   21 ar
100 | 44   21   22 ar
101 | 45   21   45 1
102 | 46   22   23 ar
103 | 47   22   46 1
104 | 48   23   24 ar
105 | 49   23   47 1
106 | 50   24   25 ar
107 | 51   24   48 1
108 | 52   25   26 1


--------------------------------------------------------------------------------
/tests.py:
--------------------------------------------------------------------------------
 1 | import os
 2 | import unittest
 3 | from prepare_ligands import convert_to_pdbqt
 4 | 
 5 | 
 6 | class LigandPrepTests(unittest.TestCase):
 7 |     
 8 |     def test_babel(self):
 9 |         mol2 = open('test_files/ZINC67842136.mol2', 'r').read()
10 |         ligand_paths = convert_to_pdbqt(mol2, 'test_files', remove_input=False)
11 |         pdbqt = ligand_paths['pdbqt']
12 |         pdbqt_contents = open(pdbqt, 'r').read()
13 |         os.remove(pdbqt)
14 |         self.assertTrue(len(pdbqt_contents) > 0)
15 |         
16 | 
17 | if __name__ == '__main__':
18 |     unittest.main()


--------------------------------------------------------------------------------