├── README
├── package.xml
└── src
    └── uk
        └── ac
            └── ebi
                └── gnx
                    ├── ParseFasta.java
                    └── CalculateGNx.java


/README:
--------------------------------------------------------------------------------
 1 | Basic Genome assembly statistic tool to calculate Nx values e.g. N50, N10, NG50
 2 | 
 3 | INSTALLATION
 4 | 
 5 | a) Download git repository
 6 | b) Compile java classes
 7 |    # in gnx-tools folder run
 8 |    mkdir bin
 9 |    javac -d bin/ src/uk/ac/ebi/gnx/*
10 | c) package jar file
11 |    # in gnx-tools foder run
12 |    ant -f package.xml
13 | d) execute jar file
14 |    java -jar gnx.jar
15 | 
16 | DONE
17 | 
18 | 
19 | 


--------------------------------------------------------------------------------
/package.xml:
--------------------------------------------------------------------------------
 1 | <?xml version="1.0" encoding="UTF-8" standalone="no"?>
 2 | <project default="create_run_jar" name="Create Runnable Jar for Project gnx-tools">
 3 |     <!--this file was created by Eclipse Runnable JAR Export Wizard-->
 4 | 	<!-- javac -d bin/ src/uk/ac/ebi/gnx/*  -->
 5 |     <!--ANT 1.7 is required                                        -->
 6 |     <target name="create_run_jar">
 7 |         <jar destfile="gnx.jar" filesetmanifest="mergewithoutmain">
 8 |             <manifest>
 9 |                 <attribute name="Main-Class" value="uk.ac.ebi.gnx.CalculateGNx"/>
10 |                 <attribute name="Class-Path" value="."/>
11 |             </manifest>
12 |             <fileset dir="bin"/>
13 |         </jar>
14 |     </target>
15 | </project>
16 | 


--------------------------------------------------------------------------------
/src/uk/ac/ebi/gnx/ParseFasta.java:
--------------------------------------------------------------------------------
  1 | /**
  2 |  * File: ParseFasta.java
  3 |  * Created by: mhaimel
  4 |  * Created on: Feb 23, 2012
  5 |  * CVS:  $Id: ParseFasta.java 1.0 Feb 23, 2012 2:29:04 PM mhaimel Exp $
  6 |  */
  7 | package uk.ac.ebi.gnx;
  8 | 
  9 | import java.io.IOException;
 10 | import java.io.InputStream;
 11 | import java.util.ArrayList;
 12 | import java.util.List;
 13 | 
 14 | /**
 15 |  * @author mhaimel
 16 |  *
 17 |  */
 18 | public class ParseFasta {
 19 | 	private volatile InputStream in = null;
 20 | 	
 21 | 	private volatile List<Integer> len = new ArrayList<Integer>();
 22 | 	private volatile long nsCnt = 0;
 23 | 	private volatile long totalCnt = 0;
 24 | 
 25 | 	private long minLen = 0;
 26 | 
 27 | 	public ParseFasta() {
 28 | 		// empty
 29 | 	}
 30 | 	public long getNsCnt() {
 31 | 		return nsCnt;
 32 | 	}	
 33 | 	public void setNsCnt(long nsCnt) {
 34 | 		this.nsCnt = nsCnt;
 35 | 	}
 36 | 	public long getTotalCnt() {
 37 | 		return totalCnt;
 38 | 	}
 39 | 	public void setTotalCnt(long totalCnt) {
 40 | 		this.totalCnt = totalCnt;
 41 | 	}
 42 | 	public void setMinLen(long minLen) {
 43 | 		this.minLen = minLen;
 44 | 	}
 45 | 	public InputStream getIn() {
 46 | 		return in;
 47 | 	}
 48 | 	public void setIn(InputStream in) {
 49 | 		this.in = in;
 50 | 	}
 51 | 	public void setLen(List<Integer> len) {
 52 | 		this.len = len;
 53 | 	}
 54 | 	public List<Integer> getLen() {
 55 | 		return len;
 56 | 	}
 57 | 	public long getMinLen() {
 58 | 		return minLen;
 59 | 	}
 60 | 
 61 | 	public void process() throws IOException {
 62 | 		int currSeqCnt = 0;
 63 | 		int currNsCnt = 0;
 64 | 		int currSeqOfOddChars = 0;
 65 | 		while(true){
 66 | 			switch (in.read()) {
 67 | 			case -1:
 68 | 				// EOF
 69 | 			case '>':
 70 | 				if(currSeqCnt > 0){
 71 | 					if(addLength(currSeqCnt)){
 72 | 						this.nsCnt += currNsCnt;
 73 | 					}
 74 | 				}
 75 | 				currSeqCnt = 0;
 76 | 				currSeqOfOddChars = 0;
 77 | 				currNsCnt = 0;
 78 | 				boolean search = true; // ugly, but should work for the moment;
 79 | 				while(search){
 80 | 					switch (in.read()) {
 81 | 					case -1:
 82 | 						// EOF
 83 | 						return;
 84 | 					case '\n':
 85 | 						search = false;
 86 | 						break;
 87 | 					default:
 88 | 						break;
 89 | 					}
 90 | 				}
 91 | 				break;
 92 | 			case '\r':
 93 | 			case '\n':
 94 | 				// ignore
 95 | 				break;
 96 | 				
 97 | 			case 'n':
 98 | 			case 'N':
 99 | 				++currNsCnt;
100 | 			case 'a':
101 | 			case 'A':
102 | 			case 't':
103 | 			case 'T':
104 | 			case 'g':
105 | 			case 'G':
106 | 			case 'c':
107 | 			case 'C':
108 | 				++ currSeqCnt;
109 | 				break;
110 | 			default:
111 | 				++currSeqOfOddChars;
112 | 				break;
113 | 			}		
114 | 		}
115 | 	}
116 | 
117 | 	private boolean addLength(int seqLen) {
118 | 		if(seqLen >= this.minLen ){
119 | 			this.totalCnt += seqLen;
120 | 			this.len.add(seqLen);
121 | 			return true;
122 | 		}
123 | 		return false;
124 | 	}
125 | 
126 | 	public void reset() {
127 | 		this.nsCnt = 0;
128 | 		this.totalCnt = 0;
129 | 		this.len.clear();
130 | 		this.in = null;
131 | 	}
132 | }
133 | 


--------------------------------------------------------------------------------
/src/uk/ac/ebi/gnx/CalculateGNx.java:
--------------------------------------------------------------------------------
  1 | /**
  2 |  * File: CalculateGNx.java
  3 |  * Created by: mhaimel
  4 |  * Created on: Feb 22, 2012
  5 |  * CVS:  $Id: CalculateGNx.java 1.0 Feb 22, 2012 4:40:13 PM mhaimel Exp $
  6 |  */
  7 | package uk.ac.ebi.gnx;
  8 | 
  9 | import java.io.BufferedInputStream;
 10 | import java.io.File;
 11 | import java.io.FileInputStream;
 12 | import java.io.IOException;
 13 | import java.io.InputStream;
 14 | import java.io.PrintStream;
 15 | import java.util.ArrayList;
 16 | import java.util.Collections;
 17 | import java.util.List;
 18 | import java.util.zip.GZIPInputStream;
 19 | 
 20 | /**
 21 |  * @author mhaimel
 22 |  *
 23 |  */
 24 | public class CalculateGNx {	
 25 | 	
 26 | 	private final double nxPosition; // default
 27 | 	
 28 | 	public CalculateGNx() {
 29 | 		this(0.5);
 30 | 	}
 31 | 
 32 | 	public CalculateGNx(double position) {
 33 | 		this.nxPosition = position;
 34 | 	}
 35 | 	
 36 | 	private void printResults(PrintStream out, List<Integer> revSortedList, long total) {
 37 | 		int tCnt = revSortedList.size();
 38 | 		long tSum = total;
 39 | //		for(Integer i : revSortedList){
 40 | //			tSum += i;
 41 | //		}
 42 | 		double cOff = ((double)tSum)*this.nxPosition;
 43 | 		
 44 | 		long sum = 0;
 45 | 		int i = 0;
 46 | 		Integer sLen = 0;
 47 | 		for(i = 0; i < tCnt; ++i){
 48 | 			sLen = revSortedList.get(i);
 49 | 			sum += sLen;
 50 | 			if(sum >= cOff){
 51 | 				break;
 52 | 			}
 53 | 		}
 54 | 		out.println(
 55 | 				"N"
 56 | 				+Double.valueOf(this.nxPosition * 100).intValue()+":\t" 
 57 | 				+ sLen
 58 | 				+ "\t("+(i+1)+" sequences)" 
 59 | 				+ "\t("+sum +" bp combined)");
 60 | 	}
 61 | 
 62 | 
 63 | 	private static void printHelp(PrintStream out) {
 64 | 		out.println("gnx [-min <val>] [-nx 25,50,75] [-g <genomeSize>] <list of fasta files>");
 65 | 		out.println("-min   Minimum bp length of a sequence to be considered");
 66 | 		out.println("-nx    Nx values to be printed seperated by ',' e.g. 50 for N50, 25 for N25");
 67 | 		out.println("-g     genome size to be used to calculte Nx values");
 68 | 		out.println("<list of fasta files>  ");
 69 | 		out.println("     o /path/to/file.fa");
 70 | 		out.println("     o use '-' for standard input");
 71 | 		out.println("     o file-a.fa file-b.fa for a list of files");
 72 | 	}
 73 | 
 74 | 	/**
 75 | 	 * @param args
 76 | 	 * @throws IOException 
 77 | 	 */
 78 | 	public static void main(String[] args) throws IOException {
 79 | 		if(args.length == 0){
 80 | 			System.err.println("Please provide an input!");
 81 | 			printHelp(System.err);
 82 | 			fail("Please provide an input!");
 83 | 		}
 84 | 		
 85 | 		List<CalculateGNx> nxList = new ArrayList<CalculateGNx>();
 86 | 		nxList.add(new CalculateGNx());
 87 | 		ParseFasta pfa = new ParseFasta();
 88 | 		long genomeSize = -1;
 89 | 		List<File> list = new ArrayList<File>();
 90 | 		for(int i = 0; i < args.length; ++i){
 91 | 			String s = args[i];
 92 | 			if(s.equals("-min")){
 93 | 				pfa.setMinLen(Long.valueOf(args[++i]));
 94 | 			} else if(s.equals("-g")){
 95 | 				genomeSize = Long.valueOf(args[++i]);
 96 | 			} else if(s.equals("-nx")){
 97 | 				nxList.clear();
 98 | 				for(String v : args[++i].split(",")){
 99 | 					if(v.length() > 0){
100 | 						nxList.add(new CalculateGNx(Double.valueOf(v)/100));
101 | 					}
102 | 				}
103 | 			} else if(s.equals("-")){
104 | 				list.add(new File("-"));
105 | 			} else {
106 | 				File f = new File(s);
107 | 				if(!f.isFile()){
108 | 					fail(s + " is not a file!!!");
109 | 				} else if(!f.canRead()){
110 | 					fail(s + " is not Readable!!!");
111 | 				}
112 | 				list.add(f);
113 | 			}
114 | 		}		
115 | 		boolean isStream = false;
116 | 		long gSize = 0;
117 | 		for(File f : list){
118 | 			try{
119 | 				if(f.getName().equals("-")){
120 | 					isStream = true;
121 | 					pfa.setIn(System.in);
122 | 				} else {
123 | 					pfa.setIn(openFile(f));
124 | 				}
125 | 				pfa.process();
126 | 				// TODO process results
127 | 				System.out.println("Results for " + f );
128 | 				System.out.println("Total number of sequences:          " + pfa.getLen().size());
129 | 				System.out.println("Total length of sequences:          " + pfa.getTotalCnt() + " bp");
130 | 				// Sort
131 | 				Collections.sort(pfa.getLen());
132 | 				System.out.println("Shortest sequence length :          " + (pfa.getLen().isEmpty()?0:pfa.getLen().get(0)) + " bp");
133 | 				
134 | 				// Reverse
135 | 				Collections.reverse(pfa.getLen());
136 | 				System.out.println("Longest sequence length  :          " + (pfa.getLen().isEmpty()?0:pfa.getLen().get(0))+ " bp");
137 | 				
138 | 				gSize = 0;
139 | 				if(genomeSize < 0){
140 | 					gSize = pfa.getTotalCnt();
141 | 				} else {
142 | 					gSize = genomeSize;
143 | 					System.out.println("-> with a provided genome size of:  " + gSize + " bp");
144 | 				}
145 | 				System.out.println("Total number of Ns in sequences:    " + pfa.getNsCnt());
146 | 				for(CalculateGNx nx : nxList){
147 | 					nx.printResults(System.out,pfa.getLen(), gSize);
148 | 				}
149 | 				if(!isStream){
150 | 					pfa.getIn().close();
151 | 				}
152 | 				pfa.setIn(null);
153 | 			} finally{
154 | 				if(pfa.getIn() != null){
155 | 					try{
156 | 						pfa.getIn().close();
157 | 					} catch (Exception e) {
158 | 						// ignore
159 | 					}
160 | 					pfa.setIn(null);
161 | 				}
162 | 			}
163 | 			System.out.println("");
164 | 			pfa.reset();
165 | 		}
166 | 	}
167 | 
168 | 	private static InputStream openFile(File f) throws IOException {
169 | 		InputStream in = new FileInputStream(f);
170 | 		if(f.getName().endsWith(".gz") || f.getName().endsWith(".gzip")){
171 | 			in = new GZIPInputStream(in);
172 | 		}
173 | 		in = new BufferedInputStream(in);
174 | 		return in;
175 | 	}
176 | 
177 | 	private static void fail(String msg) {
178 | 		System.err.println(msg);
179 | 		System.exit(1);
180 | 	}
181 | 	
182 | 
183 | 
184 | 
185 | }
186 | 


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