add option bwa index...todo: a-stats

mdcao · Aug 8, 2017 · 64334a5 · 64334a5
1 parent 3f37020
commit 64334a5
Show file tree

Hide file tree

Showing 5 changed files with 59 additions and 30 deletions.
diff --git a/src/dev/java/japsadev/bio/hts/newscarf/GraphExplore.java b/src/dev/java/japsadev/bio/hts/newscarf/GraphExplore.java
@@ -26,8 +26,10 @@ public static void main(String args[]) {
 
     public GraphExplore() throws IOException{
     	//System.setProperty("org.graphstream.ui.renderer", "org.graphstream.ui.j2dviewer.J2DGraphRenderer"); 
-    	String sample="EcK12S-careful";
+//    	String sample="EcK12S-careful";
 //    	String sample="Kp2146-careful";
+//    	String sample="meta-careful";
+    	String sample="cp_S5";
 
         HybridAssembler ass = new HybridAssembler(spadesFolder+sample+"/assembly_graph.fastg");
     	BidirectedGraph graph= ass.simGraph;
@@ -43,8 +45,8 @@ public GraphExplore() throws IOException{
 
 
         for (Node node : graph) {
-            node.addAttribute("ui.label", node.getId());
-            node.setAttribute("ui.style", "text-offset: -10;"); 
+//            node.addAttribute("ui.label", node.getId());
+//            node.setAttribute("ui.style", "text-offset: -10;"); 
             if(BidirectedGraph.isUnique(node))
             	node.setAttribute("ui.class", "marked");
         }
@@ -57,7 +59,7 @@ public GraphExplore() throws IOException{
         try {
         	HybridAssembler.promptEnterKey();
 			ass.reduceFromSPAdesPaths(spadesFolder+sample+"/contigs.paths");
-//			HybridAssembler.promptEnterKey();
+			HybridAssembler.promptEnterKey();
 			ass.assembly(spadesFolder+sample+"/assembly_graph.sam");
 
 		} catch (IOException e) {

diff --git a/src/dev/java/japsadev/bio/hts/newscarf/HybridAssembler.java b/src/dev/java/japsadev/bio/hts/newscarf/HybridAssembler.java
@@ -74,7 +74,8 @@ public void assembly(String bamFile) throws IOException{
 
 				if(p!=null)
 					System.out.println("Final path found: " + p.getId());
-				reduce2(p);
+				reduce(p);
+//				reduce2(p);
 				samList = new ArrayList<Alignment>();
 				//readID = myRec.readID;	
 			}	
@@ -102,7 +103,8 @@ public void reduceFromSPAdesPaths(String paths) throws IOException{
 			if(s.contains("NODE")){
 				if(flag){
 					BidirectedPath path=new BidirectedPath(simGraph, curpath);
-			    	reduce2(path);
+			    	reduce(path);
+//			    	reduce2(path);
 				}
 				flag=s.contains("'")?false:true;
 				curpath=new String();

diff --git a/src/dev/java/japsadev/tools/GapCloserCmd.java b/src/dev/java/japsadev/tools/GapCloserCmd.java
@@ -70,7 +70,7 @@ public GapCloserCmd(){
 		addString("seqFile", null, "Name of the assembly file (sorted by length)",true);
 
 		addString("input", "-", "Name of the input file, - for stdin", true);
-		addString("format", "sam", "format of the input fastq/fasta or sam/bam");
+		addString("format", "sam", "Format of the input: fastq/fasta or sam/bam", true);
 		addBoolean("index", true, "Whether to index the contigs sequence by the aligner or not.");
 
 		addString("bwaExe", "bwa", "Path to bwa");
@@ -153,21 +153,20 @@ public static void main(String[] args) throws IOException, InterruptedException
 				ProcessBuilder pb = new ProcessBuilder(bwaExe).redirectErrorStream(true);
 				Process process =  pb.start();
 				BufferedReader bf = SequenceReader.openFile(process.getInputStream());
-				if(process.waitFor()!=0){
-					System.err.println("bwa failed!");//???
-					System.exit(1);
-				}
+
+
 				String line;
 				String version = "";
 				Pattern versionPattern = Pattern.compile("^Version:\\s(\\d+\\.\\d+\\.\\d+).*");
+				Matcher matcher=versionPattern.matcher("");
+
 				while ((line = bf.readLine())!=null){				
-					System.out.println(line);
-					Matcher matcher =versionPattern.matcher(line);
+					matcher.reset(line);
 					if (matcher.find()){
-						System.out.print(line);
 					    version = matcher.group(1);
+					    break;//while
 					}
-					break;//while
+
 
 				}	
 				bf.close();
@@ -187,10 +186,7 @@ public static void main(String[] args) throws IOException, InterruptedException
 				if(cmdLine.getBooleanVal("index")){
 					ProcessBuilder pb2 = new ProcessBuilder(bwaExe,"index",sequenceFile);
 					Process indexProcess =  pb2.start();
-					if(indexProcess.waitFor()!=0){
-						System.err.println("Indexing failed!");
-						System.exit(1);
-					}
+					indexProcess.waitFor();
 				}
 			}catch (IOException e){
 				System.err.println(e.getMessage());

diff --git a/src/main/java/japsa/bio/hts/scaffold/AlignmentRecord.java b/src/main/java/japsa/bio/hts/scaffold/AlignmentRecord.java
@@ -83,6 +83,13 @@ public AlignmentRecord(String readID, int refStart, int refEnd, int readLength,
 		this.score = score;
 	}
 	public AlignmentRecord(SAMRecord sam, Contig ctg) {
+		if(!sam.getReferenceName().equals(ctg.getName())){
+			System.err.println("Reference in SAM file doesn't agree with contigs name: "
+							+ sam.getReferenceName() + " != " + ctg.getName());
+			System.err.println("Hint: SAM file must resulted from alignment between long reads and contigs!");
+			System.exit(1);
+		}
+
 //		readID = Integer.parseInt(sam.getReadName().split("_")[0]);
 		readID = sam.getReadName();
 

diff --git a/src/main/java/japsa/tools/bio/np/GapCloserCmd.java b/src/main/java/japsa/tools/bio/np/GapCloserCmd.java
@@ -47,6 +47,8 @@
 import java.io.BufferedReader;
 import java.io.File;
 import java.io.IOException;
+import java.util.regex.Matcher;
+import java.util.regex.Pattern;
 
 /**
  * @author sonnguyen, minhduc
@@ -67,10 +69,11 @@ public GapCloserCmd(){
 		setDesc(annotation.scriptDesc());
 
 		addString("seqFile", null, "Name of the assembly file (sorted by length)",true);
-		//addString("bamFile", null, "Name of the bam file", true);
 
 		addString("input", "-", "Name of the input file, - for stdin", true);
-		addString("format", "fastq/fasta", "format of the input fastq/fasta or sam/bam");
+		addString("format", "sam", "Format of the input: fastq/fasta or sam/bam", true);
+		addBoolean("index", true, "Whether to index the contigs sequence by the aligner or not.");
+
 		addString("bwaExe", "bwa", "Path to bwa");
 		addInt("bwaThread", 4, "Theads used by bwa");
 		addBoolean("long", false, "Whether report all sequences, including short/repeat contigs (default) or only long/unique/completed sequences.");
@@ -135,24 +138,43 @@ public static void main(String[] args) throws
 				ProcessBuilder pb = new ProcessBuilder(bwaExe).redirectErrorStream(true);
 				Process process =  pb.start();
 				BufferedReader bf = SequenceReader.openFile(process.getInputStream());
+
+
 				String line;
 				String version = "";
-				while ((line = bf.readLine())!=null){
-					if (line.startsWith("Version: ")){
-						version = line.substring(9).trim();
-						break;//while
-					}				
+				Pattern versionPattern = Pattern.compile("^Version:\\s(\\d+\\.\\d+\\.\\d+).*");
+				Matcher matcher=versionPattern.matcher("");
+
+				while ((line = bf.readLine())!=null){				
+					matcher.reset(line);
+					if (matcher.find()){
+					    version = matcher.group(1);
+					    break;//while
+					}
+
+
 				}	
 				bf.close();
+
 				if (version.length() == 0){
-					System.err.println(bwaExe + " is not the rith path to bwa. bwa is required");
+					LOG.error(bwaExe + " is not the right path to bwa. bwa is required");
 					System.exit(1);
 				}else{
-					if (!version.startsWith("0.7.1")){
-						System.err.println(" Require bwa of 0.7.11 or above");
+					LOG.info("bwa version: " + version);
+					if (version.compareTo("0.7.11") < 0){
+						LOG.error(" Require bwa of 0.7.11 or above");
 						System.exit(1);
 					}
 				}
+
+				//run indexing 
+				if(cmdLine.getBooleanVal("index")){
+					LOG.info("bwa index running...");
+					ProcessBuilder pb2 = new ProcessBuilder(bwaExe,"index",sequenceFile);
+					Process indexProcess =  pb2.start();
+					indexProcess.waitFor();
+					LOG.info("bwa index finished!");
+				}
 			}catch (IOException e){
 				System.err.println(e.getMessage());
 				System.exit(1);
@@ -161,7 +183,7 @@ public static void main(String[] args) throws
 		}else if (format.startsWith("sam") || format.startsWith("bam")){
 			// no problem
 		}else{
-			System.err.println("I dont understand format " + format);
+			LOG.error("Unrecognized format: " + format);
 			System.exit(1);
 		}