jpayne@68: package consensus;
jpayne@68: 
jpayne@68: import java.io.PrintStream;
jpayne@68: import java.util.ArrayList;
jpayne@68: import java.util.LinkedHashMap;
jpayne@68: import java.util.Map.Entry;
jpayne@68: import java.util.concurrent.atomic.AtomicIntegerArray;
jpayne@68: import java.util.concurrent.atomic.AtomicLongArray;
jpayne@68: 
jpayne@68: import fileIO.ByteFile;
jpayne@68: import fileIO.FileFormat;
jpayne@68: import fileIO.ReadWrite;
jpayne@68: import shared.Parse;
jpayne@68: import shared.Parser;
jpayne@68: import shared.PreParser;
jpayne@68: import shared.ReadStats;
jpayne@68: import shared.Shared;
jpayne@68: import shared.Timer;
jpayne@68: import shared.Tools;
jpayne@68: import stream.ConcurrentReadInputStream;
jpayne@68: import stream.ConcurrentReadOutputStream;
jpayne@68: import stream.FastaReadInputStream;
jpayne@68: import stream.Read;
jpayne@68: import stream.SamLine;
jpayne@68: import stream.SamReadStreamer;
jpayne@68: import stream.SamStreamer;
jpayne@68: import structures.ByteBuilder;
jpayne@68: import structures.ListNum;
jpayne@68: import template.Accumulator;
jpayne@68: import template.ThreadWaiter;
jpayne@68: import var2.SamFilter;
jpayne@68: 
jpayne@68: /**
jpayne@68:  * Resizes scaffold gaps to represent the best estimate 
jpayne@68:  * based on the insert size distribution of paired reads.
jpayne@68:  * 
jpayne@68:  * @author Brian Bushnell
jpayne@68:  * @date September 11, 2019
jpayne@68:  *
jpayne@68:  */
jpayne@68: public class FixScaffoldGaps implements Accumulator<FixScaffoldGaps.ProcessThread> {
jpayne@68: 	
jpayne@68: 	/*--------------------------------------------------------------*/
jpayne@68: 	/*----------------        Initialization        ----------------*/
jpayne@68: 	/*--------------------------------------------------------------*/
jpayne@68: 	
jpayne@68: 	/**
jpayne@68: 	 * Code entrance from the command line.
jpayne@68: 	 * @param args Command line arguments
jpayne@68: 	 */
jpayne@68: 	public static void main(String[] args){
jpayne@68: 		//Start a timer immediately upon code entrance.
jpayne@68: 		Timer t=new Timer();
jpayne@68: 		
jpayne@68: 		//Create an instance of this class
jpayne@68: 		FixScaffoldGaps x=new FixScaffoldGaps(args);
jpayne@68: 		
jpayne@68: 		//Run the object
jpayne@68: 		x.process(t);
jpayne@68: 		
jpayne@68: 		//Close the print stream if it was redirected
jpayne@68: 		Shared.closeStream(x.outstream);
jpayne@68: 	}
jpayne@68: 	
jpayne@68: 	/**
jpayne@68: 	 * Constructor.
jpayne@68: 	 * @param args Command line arguments
jpayne@68: 	 */
jpayne@68: 	public FixScaffoldGaps(String[] args){
jpayne@68: 		
jpayne@68: 		{//Preparse block for help, config files, and outstream
jpayne@68: 			PreParser pp=new PreParser(args, getClass(), false);
jpayne@68: 			args=pp.args;
jpayne@68: 			outstream=pp.outstream;
jpayne@68: 		}
jpayne@68: 		
jpayne@68: 		//Set shared static variables prior to parsing
jpayne@68: 		ReadWrite.USE_PIGZ=ReadWrite.USE_UNPIGZ=true;
jpayne@68: 		ReadWrite.MAX_ZIP_THREADS=Shared.threads();
jpayne@68: 		
jpayne@68: 		samFilter.includeUnmapped=false;
jpayne@68: 		samFilter.includeSupplimentary=false;
jpayne@68: //		samFilter.includeDuplicate=false;
jpayne@68: 		samFilter.includeNonPrimary=false;
jpayne@68: 		samFilter.includeQfail=false;
jpayne@68: //		samFilter.minMapq=4;
jpayne@68: 		
jpayne@68: 		{//Parse the arguments
jpayne@68: 			final Parser parser=parse(args);
jpayne@68: 			
jpayne@68: 			Parser.processQuality();
jpayne@68: 			
jpayne@68: 			maxReads=parser.maxReads;
jpayne@68: 			overwrite=ReadStats.overwrite=parser.overwrite;
jpayne@68: 			append=ReadStats.append=parser.append;
jpayne@68: 			
jpayne@68: 			in=parser.in1;
jpayne@68: 			extin=parser.extin;
jpayne@68: 
jpayne@68: 			out=parser.out1;
jpayne@68: 			extout=parser.extout;
jpayne@68: 		}
jpayne@68: 		
jpayne@68: 		{
jpayne@68: //			if("auto".equalsIgnoreCase(atomic)){Scaffold.setCA3A(Shared.threads()>8);}
jpayne@68: //			else{Scaffold.setCA3A(Parse.parseBoolean(atomic));}
jpayne@68: 			samFilter.setSamtoolsFilter();
jpayne@68: 			
jpayne@68: 			streamerThreads=Tools.max(1, Tools.min(streamerThreads, Shared.threads()));
jpayne@68: 			assert(streamerThreads>0) : streamerThreads;
jpayne@68: 		}
jpayne@68: 
jpayne@68: 		validateParams();
jpayne@68: 		fixExtensions(); //Add or remove .gz or .bz2 as needed
jpayne@68: 		checkFileExistence(); //Ensure files can be read and written
jpayne@68: 		checkStatics(); //Adjust file-related static fields as needed for this program 
jpayne@68: 		
jpayne@68: 		//Create output FileFormat objects
jpayne@68: 		ffout=FileFormat.testOutput(out, FileFormat.FASTA, extout, true, overwrite, append, ordered);
jpayne@68: 
jpayne@68: 		//Create input FileFormat objects
jpayne@68: 		ffin=FileFormat.testInput(in, FileFormat.SAM, extin, true, true);
jpayne@68: 		ffref=FileFormat.testInput(ref, FileFormat.FASTA, null, true, true);
jpayne@68: 	}
jpayne@68: 	
jpayne@68: 	/*--------------------------------------------------------------*/
jpayne@68: 	/*----------------    Initialization Helpers    ----------------*/
jpayne@68: 	/*--------------------------------------------------------------*/
jpayne@68: 	
jpayne@68: 	/** Parse arguments from the command line */
jpayne@68: 	private Parser parse(String[] args){
jpayne@68: 		
jpayne@68: 		//Create a parser object
jpayne@68: 		Parser parser=new Parser();
jpayne@68: 		
jpayne@68: 		//Set any necessary Parser defaults here
jpayne@68: 		//parser.foo=bar;
jpayne@68: 		
jpayne@68: 		//Parse each argument
jpayne@68: 		for(int i=0; i<args.length; i++){
jpayne@68: 			String arg=args[i];
jpayne@68: 			
jpayne@68: 			//Break arguments into their constituent parts, in the form of "a=b"
jpayne@68: 			String[] split=arg.split("=");
jpayne@68: 			String a=split[0].toLowerCase();
jpayne@68: 			String b=split.length>1 ? split[1] : null;
jpayne@68: 			if(b!=null && b.equalsIgnoreCase("null")){b=null;}
jpayne@68: 			
jpayne@68: 			if(a.equals("verbose")){
jpayne@68: 				verbose=Parse.parseBoolean(b);
jpayne@68: 			}else if(a.equals("ref") || a.equals("scaffolds")){
jpayne@68: 				ref=b;
jpayne@68: 			}else if(a.equals("insertlist")){
jpayne@68: 				insertList=b;
jpayne@68: 			}else if(a.equals("ordered")){
jpayne@68: 				ordered=Parse.parseBoolean(b);
jpayne@68: 			}else if(a.equalsIgnoreCase("ns") || a.equalsIgnoreCase("n") || a.equalsIgnoreCase("scaffoldbreak") || a.equalsIgnoreCase("gap")){
jpayne@68: 				scaffoldBreakNs=Integer.parseInt(b);
jpayne@68: 				assert(scaffoldBreakNs>0);
jpayne@68: 			}else if(a.equalsIgnoreCase("mindepth")){
jpayne@68: 				minDepth=Integer.parseInt(b);
jpayne@68: 				assert(minDepth>0);
jpayne@68: 			}else if(a.equalsIgnoreCase("trim") || a.equalsIgnoreCase("trimFraction") || a.equalsIgnoreCase("border")){
jpayne@68: 				trimFraction=Float.parseFloat(b);
jpayne@68: 				assert(trimFraction>=0 && trimFraction<=1) : "trimFraction should be between 0 and 1";
jpayne@68: 			}else if(a.equals("clearfilters") || a.equals("clearfilter")){
jpayne@68: 				if(Parse.parseBoolean(b)){
jpayne@68: 					samFilter.clear();
jpayne@68: 				}
jpayne@68: 			}else if(a.equals("parse_flag_goes_here")){
jpayne@68: 				long fake_variable=Parse.parseKMG(b);
jpayne@68: 				//Set a variable here
jpayne@68: 			}else if(samFilter.parse(arg, a, b)){
jpayne@68: 				//do nothing
jpayne@68: 			}else if(parser.parse(arg, a, b)){//Parse standard flags in the parser
jpayne@68: 				//do nothing
jpayne@68: 			}else{
jpayne@68: 				outstream.println("Unknown parameter "+args[i]);
jpayne@68: 				assert(false) : "Unknown parameter "+args[i];
jpayne@68: 			}
jpayne@68: 		}
jpayne@68: 		
jpayne@68: 		return parser;
jpayne@68: 	}
jpayne@68: 	
jpayne@68: 	/** Add or remove .gz or .bz2 as needed */
jpayne@68: 	private void fixExtensions(){
jpayne@68: 		in=Tools.fixExtension(in);
jpayne@68: 		ref=Tools.fixExtension(ref);
jpayne@68: 	}
jpayne@68: 	
jpayne@68: 	/** Ensure files can be read and written */
jpayne@68: 	private void checkFileExistence(){
jpayne@68: 
jpayne@68: 		//Ensure there is an input file
jpayne@68: 		if(in==null){throw new RuntimeException("Error - an input file is required.");}
jpayne@68: 
jpayne@68: 		//Ensure there is an input file
jpayne@68: 		if(ref==null){throw new RuntimeException("Error - a reference file is required.");}
jpayne@68: 		
jpayne@68: 		//Ensure output files can be written
jpayne@68: 		if(!Tools.testOutputFiles(overwrite, append, false, out)){
jpayne@68: 			outstream.println((out==null)+", "+out);
jpayne@68: 			throw new RuntimeException("\n\noverwrite="+overwrite+"; Can't write to output file "+out+"\n");
jpayne@68: 		}
jpayne@68: 		
jpayne@68: 		//Ensure input files can be read
jpayne@68: 		if(!Tools.testInputFiles(false, true, in, ref)){
jpayne@68: 			throw new RuntimeException("\nCan't read some input files.\n");  
jpayne@68: 		}
jpayne@68: 		
jpayne@68: 		//Ensure that no file was specified multiple times
jpayne@68: 		if(!Tools.testForDuplicateFiles(true, in, ref, out)){
jpayne@68: 			throw new RuntimeException("\nSome file names were specified multiple times.\n");
jpayne@68: 		}
jpayne@68: 	}
jpayne@68: 	
jpayne@68: 	/** Adjust file-related static fields as needed for this program */
jpayne@68: 	private static void checkStatics(){
jpayne@68: 		//Adjust the number of threads for input file reading
jpayne@68: 		if(!ByteFile.FORCE_MODE_BF1 && !ByteFile.FORCE_MODE_BF2 && Shared.threads()>2){
jpayne@68: 			ByteFile.FORCE_MODE_BF2=true;
jpayne@68: 		}
jpayne@68: 		
jpayne@68: 		assert(FastaReadInputStream.settingsOK());
jpayne@68: 	}
jpayne@68: 	
jpayne@68: 	/** Ensure parameter ranges are within bounds and required parameters are set */
jpayne@68: 	private boolean validateParams(){
jpayne@68: //		assert(minfoo>0 && minfoo<=maxfoo) : minfoo+", "+maxfoo;
jpayne@68: 		return true;
jpayne@68: 	}
jpayne@68: 	
jpayne@68: 	/*--------------------------------------------------------------*/
jpayne@68: 	/*----------------         Outer Methods        ----------------*/
jpayne@68: 	/*--------------------------------------------------------------*/
jpayne@68: 
jpayne@68: 	/** Create read streams and process all data */
jpayne@68: 	void process(Timer t){
jpayne@68: 		
jpayne@68: 		//Turn off read validation in the input threads to increase speed
jpayne@68: 		final boolean vic=Read.VALIDATE_IN_CONSTRUCTOR;
jpayne@68: 		Read.VALIDATE_IN_CONSTRUCTOR=Shared.threads()<4;
jpayne@68: 		
jpayne@68: 		//Create a read input stream
jpayne@68: 		final SamStreamer ss=makeStreamer(ffin);
jpayne@68: 		
jpayne@68: 		//Load reference
jpayne@68: 		loadReferenceCustom();
jpayne@68: 		
jpayne@68: 		//Reset counters
jpayne@68: 		readsProcessed=readsOut=0;
jpayne@68: 		basesProcessed=basesOut=0;
jpayne@68: 		
jpayne@68: 		//Process the reads in separate threads
jpayne@68: 		spawnThreads(ss);
jpayne@68: 
jpayne@68: 		//Optionally create a read output stream
jpayne@68: 		final ConcurrentReadOutputStream ros=makeCros();
jpayne@68: 		
jpayne@68: 		if(verbose){outstream.println("Fixing reference.");}
jpayne@68: 		
jpayne@68: 		fixScaffolds(ros);
jpayne@68: 		
jpayne@68: 		if(verbose){outstream.println("Finished; closing streams.");}
jpayne@68: 		
jpayne@68: 		//Write anything that was accumulated by ReadStats
jpayne@68: 		errorState|=ReadStats.writeAll();
jpayne@68: 		//Close the read streams
jpayne@68: 		errorState|=ReadWrite.closeStream(ros);
jpayne@68: 		
jpayne@68: 		//Reset read validation
jpayne@68: 		Read.VALIDATE_IN_CONSTRUCTOR=vic;
jpayne@68: 		
jpayne@68: 		//Report timing and results
jpayne@68: 		t.stop();
jpayne@68: 		outstream.println(Tools.timeReadsBasesProcessed(t, readsProcessed, basesProcessed, 8));
jpayne@68: 		outstream.println(Tools.readsBasesOut(readsProcessed, basesProcessed, scaffoldsOut, scaffoldLengthOut, 8, false));
jpayne@68: 		
jpayne@68: 		outstream.println();
jpayne@68: 		outstream.println(Tools.number("Average Insert", totalAverageInsert, 2, 8));
jpayne@68: 		outstream.println(Tools.number("Gaps Unchanged", gapsUnchanged, 8));
jpayne@68: 		outstream.println(Tools.number("Gaps Widened  ", gapsWidened, 8));
jpayne@68: 		outstream.println(Tools.number("Gaps Narrowed ", gapsNarrowed, 8));
jpayne@68: 		outstream.println(Tools.number("Ns Total      ", nsTotal, 8));
jpayne@68: 		outstream.println(Tools.number("Ns Added      ", nsAdded, 8));
jpayne@68: 		outstream.println(Tools.number("Ns Removed    ", nsRemoved, 8));
jpayne@68: 		
jpayne@68: 		
jpayne@68: 		//Throw an exception of there was an error in a thread
jpayne@68: 		if(errorState){
jpayne@68: 			throw new RuntimeException(getClass().getName()+" terminated in an error state; the output may be corrupt.");
jpayne@68: 		}
jpayne@68: 	}
jpayne@68: 
jpayne@68: 	private synchronized void loadReferenceCustom(){
jpayne@68: 		assert(!loadedRef);
jpayne@68: 		ConcurrentReadInputStream cris=makeRefCris();
jpayne@68: 		for(ListNum<Read> ln=cris.nextList(); ln!=null && ln.size()>0; ln=cris.nextList()) {
jpayne@68: 			for(Read r : ln){
jpayne@68: 				String name=r.id;
jpayne@68: 				String name2=Tools.trimToWhitespace(r.id);
jpayne@68: 				Scaffold scaf=new Scaffold(name, r.bases, r.numericID);
jpayne@68: 				refMap.put(name, scaf);
jpayne@68: 				refMap2.put(name2, scaf);
jpayne@68: 			}
jpayne@68: 		}
jpayne@68: 		loadedRef=true;
jpayne@68: 	}
jpayne@68: 	
jpayne@68: 	private ConcurrentReadInputStream makeRefCris(){
jpayne@68: 		ConcurrentReadInputStream cris=ConcurrentReadInputStream.getReadInputStream(maxReads, true, ffref, null);
jpayne@68: 		cris.start(); //Start the stream
jpayne@68: 		if(verbose){outstream.println("Started cris");}
jpayne@68: 		boolean paired=cris.paired();
jpayne@68: 		assert(!paired) : "References should not be paired.";
jpayne@68: 		return cris;
jpayne@68: 	}
jpayne@68: 	
jpayne@68: 	private SamStreamer makeStreamer(FileFormat ff){
jpayne@68: 		if(ff==null){return null;}
jpayne@68: 		SamStreamer ss=new SamReadStreamer(ff, streamerThreads, true, maxReads);
jpayne@68: 		ss.start(); //Start the stream
jpayne@68: 		if(verbose){outstream.println("Started Streamer");}
jpayne@68: 		return ss;
jpayne@68: 	}
jpayne@68: 	
jpayne@68: 	private ConcurrentReadOutputStream makeCros(){
jpayne@68: 		if(ffout==null){return null;}
jpayne@68: 
jpayne@68: 		//Select output buffer size based on whether it needs to be ordered
jpayne@68: 		final int buff=(ordered ? Tools.mid(16, 128, (Shared.threads()*2)/3) : 8);
jpayne@68: 
jpayne@68: 		final ConcurrentReadOutputStream ros=ConcurrentReadOutputStream.getStream(ffout, null, buff, null, false);
jpayne@68: 		ros.start(); //Start the stream
jpayne@68: 		return ros;
jpayne@68: 	}
jpayne@68: 	
jpayne@68: 	/*--------------------------------------------------------------*/
jpayne@68: 	/*----------------       Thread Management      ----------------*/
jpayne@68: 	/*--------------------------------------------------------------*/
jpayne@68: 	
jpayne@68: 	/** Spawn process threads */
jpayne@68: 	private void spawnThreads(final SamStreamer ss){
jpayne@68: 		
jpayne@68: 		//Do anything necessary prior to processing
jpayne@68: 		
jpayne@68: 		//Determine how many threads may be used
jpayne@68: 		final int threads=Shared.threads();
jpayne@68: 		
jpayne@68: 		//Fill a list with ProcessThreads
jpayne@68: 		ArrayList<ProcessThread> alpt=new ArrayList<ProcessThread>(threads);
jpayne@68: 		for(int i=0; i<threads; i++){
jpayne@68: 			alpt.add(new ProcessThread(ss, i));
jpayne@68: 		}
jpayne@68: 		
jpayne@68: 		//Start the threads
jpayne@68: 		for(ProcessThread pt : alpt){
jpayne@68: 			pt.start();
jpayne@68: 		}
jpayne@68: 		
jpayne@68: 		//Wait for threads to finish
jpayne@68: 		boolean success=ThreadWaiter.waitForThreads(alpt, this);
jpayne@68: 		errorState&=!success;
jpayne@68: 		
jpayne@68: 		//Do anything necessary after processing
jpayne@68: 		totalAverageInsert=totalInsertSum/(double)totalInsertCount;
jpayne@68: 		insertByPercentile=Tools.makeHistogram(insertCounts, buckets);
jpayne@68: 	}
jpayne@68: 	
jpayne@68: 	@Override
jpayne@68: 	public final void accumulate(ProcessThread pt){
jpayne@68: 		readsProcessed+=pt.readsProcessedT;
jpayne@68: 		basesProcessed+=pt.basesProcessedT;
jpayne@68: 		readsOut+=pt.readsOutT;
jpayne@68: 		basesOut+=pt.basesOutT;
jpayne@68: 		
jpayne@68: 		totalInsertSum+=pt.totalInsertSumT;
jpayne@68: 		totalInsertCount+=pt.totalInsertCountT;
jpayne@68: 		
jpayne@68: 		errorState|=(!pt.success);
jpayne@68: 	}
jpayne@68: 	
jpayne@68: 	@Override
jpayne@68: 	public final boolean success(){return !errorState;}
jpayne@68: 	
jpayne@68: 	/*--------------------------------------------------------------*/
jpayne@68: 	/*----------------         Inner Methods        ----------------*/
jpayne@68: 	/*--------------------------------------------------------------*/
jpayne@68: 	
jpayne@68: 	private void fixScaffolds(ConcurrentReadOutputStream ros){
jpayne@68: 		ByteBuilder bb=new ByteBuilder(1000000);
jpayne@68: 
jpayne@68: 		ArrayList<Read> list=new ArrayList<Read>(200);
jpayne@68: 		long num=0;
jpayne@68: 		long lengthSum=0;
jpayne@68: 		for(Entry<String, Scaffold> e : refMap.entrySet()){
jpayne@68: 			Scaffold scaf=e.getValue();
jpayne@68: 			Read r=scaf.fixScaffold(bb);
jpayne@68: 			lengthSum+=r.length();
jpayne@68: 			list.add(r);
jpayne@68: 			scaffoldsOut++;
jpayne@68: 			scaffoldLengthOut+=r.length();
jpayne@68: 			
jpayne@68: 			if(list.size()>=200 || lengthSum>=100000){
jpayne@68: 				if(ros!=null){ros.add(list, num);}
jpayne@68: 				list=new ArrayList<Read>(200);
jpayne@68: 				num++;
jpayne@68: 				lengthSum=0;
jpayne@68: 			}
jpayne@68: 		}
jpayne@68: 		if(list.size()>0){
jpayne@68: 			if(ros!=null){ros.add(list, num);}
jpayne@68: 		}
jpayne@68: 	}
jpayne@68: 	
jpayne@68: 	private static int calcInsertSize(SamLine sl) {
jpayne@68: 		assert(sl.mapped() && sl.pairedOnSameChrom());
jpayne@68: 		assert(sl.primary());
jpayne@68: 		assert(!sl.supplementary());
jpayne@68: 		assert(sl.leftmost());
jpayne@68: 		
jpayne@68: 		assert(sl.tlen>0) : sl.tlen+"\n\n"+sl;
jpayne@68: 		return sl.tlen>0 ? sl.tlen : -sl.tlen;
jpayne@68: 		
jpayne@68: //		final int insertSize;
jpayne@68: //		String insertTag=null;
jpayne@68: //		if(sl.optional!=null){
jpayne@68: //			for(String s : sl.optional){
jpayne@68: //				if(s.startsWith("X8:Z:")){
jpayne@68: //					insertTag=s;
jpayne@68: //					break;
jpayne@68: //				}
jpayne@68: //			}
jpayne@68: //		}
jpayne@68: //		if(insertTag!=null){
jpayne@68: //			insertSize=Integer.parseInt(insertTag.substring(5));
jpayne@68: //		}else{
jpayne@68: //			insertSize=sl.tlen;//This is unsafe due to indels.
jpayne@68: //			assert(false) : "Reads need insert size tags.";
jpayne@68: //		}
jpayne@68: //		assert(insertSize>0) : sl;
jpayne@68: //		
jpayne@68: //		return insertSize;
jpayne@68: 	}
jpayne@68: 	
jpayne@68: 	/*--------------------------------------------------------------*/
jpayne@68: 	/*----------------         Inner Classes        ----------------*/
jpayne@68: 	/*--------------------------------------------------------------*/
jpayne@68: 	
jpayne@68: 	/** This class is static to prevent accidental writing to shared variables.
jpayne@68: 	 * It is safe to remove the static modifier. */
jpayne@68: 	class ProcessThread extends Thread {
jpayne@68: 		
jpayne@68: 		//Constructor
jpayne@68: 		ProcessThread(final SamStreamer ss_, final int tid_){
jpayne@68: 			ss=ss_;
jpayne@68: 			tid=tid_;
jpayne@68: 		}
jpayne@68: 		
jpayne@68: 		//Called by start()
jpayne@68: 		@Override
jpayne@68: 		public void run(){
jpayne@68: 			//Do anything necessary prior to processing
jpayne@68: 			
jpayne@68: 			//Process the reads
jpayne@68: 			processInner();
jpayne@68: 			
jpayne@68: 			//Do anything necessary after processing
jpayne@68: 			
jpayne@68: 			//Indicate successful exit status
jpayne@68: 			success=true;
jpayne@68: 		}
jpayne@68: 		
jpayne@68: 		/** Iterate through the reads */
jpayne@68: 		void processInner(){
jpayne@68: 			
jpayne@68: 			//Grab and process all lists
jpayne@68: 			for(ListNum<Read> ln=ss.nextReads(); ln!=null; ln=ss.nextReads()){
jpayne@68: //				if(verbose){outstream.println("Got list of size "+list.size());} //Disabled due to non-static access
jpayne@68: 				
jpayne@68: 				processList(ln);
jpayne@68: 			}
jpayne@68: 			
jpayne@68: 		}
jpayne@68: 		
jpayne@68: 		void processList(ListNum<Read> ln){
jpayne@68: 
jpayne@68: 			//Grab the actual read list from the ListNum
jpayne@68: 			final ArrayList<Read> reads=ln.list;
jpayne@68: 			
jpayne@68: 			//Loop through each read in the list
jpayne@68: 			for(int idx=0; idx<reads.size(); idx++){
jpayne@68: 				final Read r=reads.get(idx);
jpayne@68: 				
jpayne@68: 				//Validate reads in worker threads
jpayne@68: 				if(!r.validated()){r.validate(true);}
jpayne@68: 
jpayne@68: 				//Track the initial length for statistics
jpayne@68: 				final int initialLength=r.length();
jpayne@68: 
jpayne@68: 				//Increment counters
jpayne@68: 				readsProcessedT+=r.pairCount();
jpayne@68: 				basesProcessedT+=initialLength;
jpayne@68: 				
jpayne@68: 				processRead(r);
jpayne@68: 			}
jpayne@68: 		}
jpayne@68: 		
jpayne@68: 		/**
jpayne@68: 		 * Process a read or a read pair.
jpayne@68: 		 * @param r Read 1
jpayne@68: 		 * @param r2 Read 2 (may be null)
jpayne@68: 		 * @return True if the reads should be kept, false if they should be discarded.
jpayne@68: 		 */
jpayne@68: 		void processRead(final Read r){
jpayne@68: 			final SamLine sl=r.samline;
jpayne@68: 			assert(sl!=null) : sl;
jpayne@68: 			if(samFilter!=null && !samFilter.passesFilter(sl)){return;}
jpayne@68: 			
jpayne@68: 			//sl.nextMapped();
jpayne@68: 			if(sl.mapped() && sl.pairedOnSameChrom() && sl.properPair() && sl.primary() && !sl.supplementary() && sl.leftmost()){
jpayne@68: 				final String rname=sl.rnameS();
jpayne@68: 				Scaffold scaf=refMap.get(rname);
jpayne@68: 				if(scaf==null){scaf=refMap2.get(Tools.trimToWhitespace(rname));}
jpayne@68: 				assert(scaf!=null) : "Can't find graph for "+rname;
jpayne@68: 				
jpayne@68: 				if(scaf!=null){
jpayne@68: 					final int insertSize=calcInsertSize(sl);
jpayne@68: 					insertCounts.incrementAndGet(Tools.mid(0, insertSize, insertCounts.length()));
jpayne@68: 					scaf.add(sl, insertSize);
jpayne@68: 
jpayne@68: 					readsOutT++;
jpayne@68: 					basesOutT+=r.length();
jpayne@68: 
jpayne@68: 					totalInsertSumT+=insertSize;
jpayne@68: 					totalInsertCountT++;
jpayne@68: 				}
jpayne@68: 			}
jpayne@68: 		}
jpayne@68: 
jpayne@68: 		/** Number of reads processed by this thread */
jpayne@68: 		protected long readsProcessedT=0;
jpayne@68: 		/** Number of bases processed by this thread */
jpayne@68: 		protected long basesProcessedT=0;
jpayne@68: 		
jpayne@68: 		/** Number of reads retained by this thread */
jpayne@68: 		protected long readsOutT=0;
jpayne@68: 		/** Number of bases retained by this thread */
jpayne@68: 		protected long basesOutT=0;
jpayne@68: 		
jpayne@68: 		protected long totalInsertSumT=0;
jpayne@68: 		protected long totalInsertCountT=0;
jpayne@68: 		
jpayne@68: 		long insertSum=0;
jpayne@68: 		
jpayne@68: 		/** True only if this thread has completed successfully */
jpayne@68: 		boolean success=false;
jpayne@68: 		
jpayne@68: 		/** Shared input stream */
jpayne@68: 		private final SamStreamer ss;
jpayne@68: 		/** Thread ID */
jpayne@68: 		final int tid;
jpayne@68: 	}
jpayne@68: 	
jpayne@68: 	/*--------------------------------------------------------------*/
jpayne@68: 	
jpayne@68: 	private class Scaffold {
jpayne@68: 		
jpayne@68: 		Scaffold(String name_, byte[] bases_, long numericID_){
jpayne@68: 			name=name_;
jpayne@68: 			bases=bases_;
jpayne@68: 			numericID=(int)numericID_;
jpayne@68: 			depthArray=new AtomicIntegerArray(bases.length);
jpayne@68: 			insertArray=new AtomicLongArray(bases.length);
jpayne@68: 		}
jpayne@68: 		
jpayne@68: 		void add(SamLine sl, int insertSize){
jpayne@68: 			assert(sl.mapped() && sl.pairedOnSameChrom());
jpayne@68: 			assert(sl.primary());
jpayne@68: 			assert(!sl.supplementary());
jpayne@68: 			assert(sl.leftmost());
jpayne@68: 
jpayne@68: //			final int insertSize=calcInsertSize(sl);
jpayne@68: 			
jpayne@68: 			int start=sl.pos-1;
jpayne@68: 			int stop=start+sl.tlen;
jpayne@68: 
jpayne@68: 			int trim=(int)(sl.length()*trimFraction);
jpayne@68: 			start+=trim;
jpayne@68: 			stop-=trim;
jpayne@68: 			
jpayne@68: 			for(int i=start; i<stop; i++){
jpayne@68: 				if(i>=0 && i<bases.length){
jpayne@68: 					depthArray.incrementAndGet(i);
jpayne@68: 					insertArray.addAndGet(i, insertSize);
jpayne@68: 				}
jpayne@68: 			}
jpayne@68: 			
jpayne@68: 		}
jpayne@68: 		
jpayne@68: 		Read fixScaffold(ByteBuilder bb){
jpayne@68: 			int streak=0;
jpayne@68: 			bb.clear();
jpayne@68: 			
jpayne@68: 			if(insertList!=null){
jpayne@68: 				for(int i=0; i<bases.length; i++) {
jpayne@68: 					bb.append(i).tab().append(depthArray.get(i)).tab().append(insertArray.get(i)/(Tools.max(1, depthArray.get(i)))).nl();
jpayne@68: 				}
jpayne@68: 				ReadWrite.writeString(bb, insertList, false);
jpayne@68: 				bb.clear();
jpayne@68: 			}
jpayne@68: 			
jpayne@68: 			for(int i=0; i<bases.length; i++){
jpayne@68: 				byte b=bases[i];
jpayne@68: 				if(b=='N'){
jpayne@68: 					streak++;
jpayne@68: 				}else{
jpayne@68: 					if(streak>=scaffoldBreakNs && i-streak>300 && i<bases.length-300){
jpayne@68: 						int pivot=i-streak/2-1;
jpayne@68: 						long depthSum=depthArray.get(pivot);
jpayne@68: 						long insertSum=insertArray.get(pivot);
jpayne@68: 						double avgInsert=(insertSum/(double)depthSum);
jpayne@68: 						
jpayne@68: 						int avgDepth=((depthArray.get(i-200-streak)+depthArray.get(i+200))/2);
jpayne@68: 						int percentile=(int)(buckets*Tools.max(0.5, 1.0-depthSum/(double)(avgDepth+depthSum)));
jpayne@68: 						int insertProxy=insertByPercentile[percentile];
jpayne@68: 						
jpayne@68: //						assert(false) : Arrays.toString(insertByPercentile);
jpayne@68: 						
jpayne@68: 						int dif=(int)Math.round(insertProxy-avgInsert);
jpayne@68: 						int toAdd=Tools.max(scaffoldBreakNs, streak+dif);
jpayne@68: 
jpayne@68: //						System.err.println("totalAverageInsert="+(int)totalAverageInsert+", avg="+(int)avgInsert+", dif="+dif);
jpayne@68: //						System.err.println("proxy="+insertProxy+", percentile="+percentile+", avgDepth="+(int)avgDepth+", depth="+depthSum);
jpayne@68: //						System.err.println("pivot="+pivot+", depthSum="+depthSum+", avg="+(int)avgInsert+", streak="+streak+", dif="+dif+", toAdd="+toAdd);
jpayne@68: 						
jpayne@68: 						if(dif>0){
jpayne@68: 							//TODO:  This is tricky because long-insert reads will be self-selected.
jpayne@68: 							//Should be compared to average coverage, or nearby coverage, and then consult a size distribution histogram.
jpayne@68: 							gapsWidened++;
jpayne@68: 							nsAdded+=dif;
jpayne@68: 						}else if(dif<0){
jpayne@68: 							gapsNarrowed++;
jpayne@68: 							nsRemoved-=dif;
jpayne@68: 						}else{
jpayne@68: 							gapsUnchanged++;
jpayne@68: 						}
jpayne@68: 						nsTotal+=toAdd;
jpayne@68: 						for(int j=0; j<toAdd; j++){
jpayne@68: 							bb.append('N');
jpayne@68: 						}
jpayne@68: 					}
jpayne@68: 					streak=0;
jpayne@68: 					bb.append(b);
jpayne@68: 				}
jpayne@68: 			}
jpayne@68: 			
jpayne@68: 			return new Read(bb.toBytes(), null, name, numericID);
jpayne@68: 		}
jpayne@68: 		
jpayne@68: 		final int numericID;
jpayne@68: 		final String name;
jpayne@68: 		final byte[] bases;
jpayne@68: 		final AtomicIntegerArray depthArray;
jpayne@68: 		final AtomicLongArray insertArray;
jpayne@68: 		
jpayne@68: 	}
jpayne@68: 	
jpayne@68: 	/*--------------------------------------------------------------*/
jpayne@68: 	/*----------------            Fields            ----------------*/
jpayne@68: 	/*--------------------------------------------------------------*/
jpayne@68: 
jpayne@68: 	/** Primary input file path */
jpayne@68: 	private String in=null;
jpayne@68: 	/** Secondary input file path */
jpayne@68: 	private String ref=null;
jpayne@68: 
jpayne@68: 	/** Primary output file path */
jpayne@68: 	private String out=null;
jpayne@68: 	
jpayne@68: 	/** Override input file extension */
jpayne@68: 	private String extin=null;
jpayne@68: 	/** Override output file extension */
jpayne@68: 	private String extout=null;
jpayne@68: 	
jpayne@68: 	private String insertList=null;
jpayne@68: 	
jpayne@68: 	/*--------------------------------------------------------------*/
jpayne@68: 
jpayne@68: 	/** Number of reads processed */
jpayne@68: 	protected long readsProcessed=0;
jpayne@68: 	/** Number of bases processed */
jpayne@68: 	protected long basesProcessed=0;
jpayne@68: 
jpayne@68: 	/** Number of reads retained */
jpayne@68: 	protected long readsOut=0;
jpayne@68: 	/** Number of bases retained */
jpayne@68: 	protected long basesOut=0;
jpayne@68: 
jpayne@68: 	protected long scaffoldsOut=0;
jpayne@68: 	protected long scaffoldLengthOut=0;
jpayne@68: 	
jpayne@68: 	protected long gapsUnchanged=0;
jpayne@68: 	protected long gapsWidened=0;
jpayne@68: 	protected long gapsNarrowed=0;
jpayne@68: 	protected long nsAdded=0;
jpayne@68: 	protected long nsRemoved=0;
jpayne@68: 	protected long nsTotal=0;
jpayne@68: 	
jpayne@68: 	protected long totalInsertSum=0;
jpayne@68: 	protected long totalInsertCount=0;
jpayne@68: 	protected double totalAverageInsert;
jpayne@68: 	
jpayne@68: 	protected AtomicLongArray insertCounts=new AtomicLongArray(20000);
jpayne@68: 	protected int[] insertByPercentile;
jpayne@68: 	
jpayne@68: 	/** Quit after processing this many input reads; -1 means no limit */
jpayne@68: 	private long maxReads=-1;
jpayne@68: 	
jpayne@68: 	/*--------------------------------------------------------------*/
jpayne@68: 	
jpayne@68: 	/** Threads dedicated to reading the sam file */
jpayne@68: 	private int streamerThreads=SamStreamer.DEFAULT_THREADS;
jpayne@68: 	
jpayne@68: 	private boolean loadedRef=false;
jpayne@68: 	
jpayne@68: 	private int scaffoldBreakNs=10;
jpayne@68: 	
jpayne@68: 	int buckets=1000;
jpayne@68: 	
jpayne@68: 	private int minDepth=10;
jpayne@68: 	
jpayne@68: 	private float trimFraction=0.4f;
jpayne@68: 	
jpayne@68: 	public final SamFilter samFilter=new SamFilter();
jpayne@68: 	
jpayne@68: 	/** Uses full ref names */
jpayne@68: 	public LinkedHashMap<String, Scaffold> refMap=new LinkedHashMap<String, Scaffold>();
jpayne@68: 	/** Uses truncated ref names */
jpayne@68: 	public LinkedHashMap<String, Scaffold> refMap2=new LinkedHashMap<String, Scaffold>();;
jpayne@68: 	
jpayne@68: 	/*--------------------------------------------------------------*/
jpayne@68: 	/*----------------         Final Fields         ----------------*/
jpayne@68: 	/*--------------------------------------------------------------*/
jpayne@68: 
jpayne@68: 	/** Primary input file */
jpayne@68: 	private final FileFormat ffin;
jpayne@68: 	/** Secondary input file */
jpayne@68: 	private final FileFormat ffref;
jpayne@68: 	
jpayne@68: 	/** Primary output file */
jpayne@68: 	private final FileFormat ffout;
jpayne@68: 	
jpayne@68: 	/*--------------------------------------------------------------*/
jpayne@68: 	/*----------------        Common Fields         ----------------*/
jpayne@68: 	/*--------------------------------------------------------------*/
jpayne@68: 	
jpayne@68: 	/** Print status messages to this output stream */
jpayne@68: 	private PrintStream outstream=System.err;
jpayne@68: 	/** Print verbose messages */
jpayne@68: 	public static boolean verbose=false;
jpayne@68: 	/** True if an error was encountered */
jpayne@68: 	public boolean errorState=false;
jpayne@68: 	/** Overwrite existing output files */
jpayne@68: 	private boolean overwrite=false;
jpayne@68: 	/** Append to existing output files */
jpayne@68: 	private boolean append=false;
jpayne@68: 	/** Reads are output in input order */
jpayne@68: 	private boolean ordered=false;
jpayne@68: 	
jpayne@68: }