jpayne@68: package clump;
jpayne@68: 
jpayne@68: import java.io.File;
jpayne@68: import java.io.PrintStream;
jpayne@68: import java.util.ArrayList;
jpayne@68: 
jpayne@68: import bloom.KCountArray;
jpayne@68: import bloom.ReadCounter;
jpayne@68: import fileIO.ByteFile;
jpayne@68: import fileIO.FileFormat;
jpayne@68: import fileIO.ReadWrite;
jpayne@68: import jgi.BBMerge;
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.FASTQ;
jpayne@68: import stream.FastaReadInputStream;
jpayne@68: import stream.Read;
jpayne@68: import structures.ListNum;
jpayne@68: 
jpayne@68: /**
jpayne@68:  * Reduces reads to their feature kmer.
jpayne@68:  * @author Brian Bushnell
jpayne@68:  * @date August 19, 2016
jpayne@68:  *
jpayne@68:  */
jpayne@68: public class PivotSet {
jpayne@68: 
jpayne@68: 	/*--------------------------------------------------------------*/
jpayne@68: 	/*----------------        Static Methods        ----------------*/
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: 		makeSet(args);
jpayne@68: 	}
jpayne@68: 	
jpayne@68: 	public static KCountArray makeSet(String[] args){
jpayne@68: 		final boolean pigz=ReadWrite.USE_PIGZ, unpigz=ReadWrite.USE_UNPIGZ;
jpayne@68: 		Timer t=new Timer();
jpayne@68: 		PivotSet x=new PivotSet(args);
jpayne@68: 		KCountArray kca=x.process(t, false);
jpayne@68: 		ReadWrite.USE_PIGZ=pigz;
jpayne@68: 		ReadWrite.USE_UNPIGZ=unpigz;
jpayne@68: 		
jpayne@68: 		//Close the print stream if it was redirected
jpayne@68: 		Shared.closeStream(x.outstream);
jpayne@68: 		
jpayne@68: 		return kca;
jpayne@68: 	}
jpayne@68: 
jpayne@68: 	/*--------------------------------------------------------------*/
jpayne@68: 	/*----------------        Initialization        ----------------*/
jpayne@68: 	/*--------------------------------------------------------------*/
jpayne@68: 	
jpayne@68: 	/**
jpayne@68: 	 * Constructor.
jpayne@68: 	 * @param args Command line arguments
jpayne@68: 	 */
jpayne@68: 	public PivotSet(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: 		ReadWrite.USE_PIGZ=ReadWrite.USE_UNPIGZ=true;
jpayne@68: 		ReadWrite.MAX_ZIP_THREADS=Shared.threads();
jpayne@68: 		
jpayne@68: 		Parser parser=new Parser();
jpayne@68: 		for(int i=0; i<args.length; i++){
jpayne@68: 			String arg=args[i];
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: 
jpayne@68: 			if(parser.parse(arg, a, b)){
jpayne@68: 				//do nothing
jpayne@68: 			}else if(a.equals("verbose")){
jpayne@68: 				verbose=KmerComparator.verbose=Parse.parseBoolean(b);
jpayne@68: 			}else if(a.equals("parse_flag_goes_here")){
jpayne@68: 				//Set a variable here
jpayne@68: 			}else if(a.equals("k")){
jpayne@68: 				k=Integer.parseInt(b);
jpayne@68: 				assert(k>0 && k<32);
jpayne@68: 			}else if(a.equals("ecco")){
jpayne@68: 				ecco=Parse.parseBoolean(b);
jpayne@68: 			}else if(a.equals("rename") || a.equals("addname")){
jpayne@68: 				//do nothing
jpayne@68: 			}else if(a.equals("rcomp") || a.equals("reversecomplement")){
jpayne@68: 				//do nothing
jpayne@68: 			}else if(a.equals("condense") || a.equals("consensus")){
jpayne@68: 				//do nothing
jpayne@68: 			}else if(a.equals("mincount") || a.equals("consensus")){
jpayne@68: 				minCount=Integer.parseInt(b);
jpayne@68: 			}else if(a.equals("correct") || a.equals("ecc")){
jpayne@68: 				//do nothing
jpayne@68: 			}else if(a.equals("groups") || a.equals("g") || a.equals("sets") || a.equals("ways")){
jpayne@68: 				//do nothing
jpayne@68: 			}else if(a.equals("seed")){
jpayne@68: 				KmerComparator.defaultSeed=Long.parseLong(b);
jpayne@68: 			}else if(a.equals("hashes")){
jpayne@68: 				KmerComparator.setHashes(Integer.parseInt(b));
jpayne@68: 			}else{
jpayne@68: 				outstream.println("Unknown parameter "+args[i]);
jpayne@68: 				assert(false) : "Unknown parameter "+args[i];
jpayne@68: 				//				throw new RuntimeException("Unknown parameter "+args[i]);
jpayne@68: 			}
jpayne@68: 		}
jpayne@68: 		
jpayne@68: 		{//Process parser fields
jpayne@68: 			Parser.processQuality();
jpayne@68: 			
jpayne@68: 			maxReads=parser.maxReads;
jpayne@68: 			
jpayne@68: 			in1=parser.in1;
jpayne@68: 			in2=parser.in2;
jpayne@68: 			
jpayne@68: 			extin=parser.extin;
jpayne@68: 		}
jpayne@68: 		
jpayne@68: 		if(in1!=null && in2==null && in1.indexOf('#')>-1 && !new File(in1).exists()){
jpayne@68: 			in2=in1.replace("#", "2");
jpayne@68: 			in1=in1.replace("#", "1");
jpayne@68: 		}
jpayne@68: 		if(in2!=null){
jpayne@68: 			if(FASTQ.FORCE_INTERLEAVED){outstream.println("Reset INTERLEAVED to false because paired input files were specified.");}
jpayne@68: 			FASTQ.FORCE_INTERLEAVED=FASTQ.TEST_INTERLEAVED=false;
jpayne@68: 		}
jpayne@68: 		
jpayne@68: 		assert(FastaReadInputStream.settingsOK());
jpayne@68: 		
jpayne@68: 		if(in1==null){throw new RuntimeException("Error - at least one input file is required.");}
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: 		ffin1=FileFormat.testInput(in1, FileFormat.FASTQ, extin, true, true);
jpayne@68: 		ffin2=FileFormat.testInput(in2, FileFormat.FASTQ, extin, true, true);
jpayne@68: 	}
jpayne@68: 	
jpayne@68: 	
jpayne@68: 	/*--------------------------------------------------------------*/
jpayne@68: 	/*----------------         Outer Methods        ----------------*/
jpayne@68: 	/*--------------------------------------------------------------*/
jpayne@68: 
jpayne@68: 	private static long getCells(double fraction, int cbits){
jpayne@68: 		final long memory=Runtime.getRuntime().maxMemory();
jpayne@68: 		final long usable=(long)Tools.max(((memory-96000000)*.73), memory*0.45);
jpayne@68: 		final double filterMem=usable*fraction;
jpayne@68: 		return (long)((filterMem*8)/cbits);
jpayne@68: 	}
jpayne@68: 	
jpayne@68: 	/** Create read streams and process all data */
jpayne@68: 	public KCountArray process(Timer t, boolean amino){
jpayne@68: 		int cbits=2;
jpayne@68: 		while((1L<<cbits)<=minCount){cbits*=2;}
jpayne@68: 		int filterHashes=2;
jpayne@68: 		float fraction=0.1f;
jpayne@68: 		long cells=getCells(fraction, cbits);
jpayne@68: 		ReadCounter rc=new ReadCounter(k, true, ecco, false, amino);
jpayne@68: 		KCountArray kca=rc.makeKca(null, null, null, cbits, cells, filterHashes, 0, maxReads, 1, 1, 1, 1, null, 0);
jpayne@68: 		
jpayne@68: 		final ConcurrentReadInputStream cris;
jpayne@68: 		{
jpayne@68: 			cris=ConcurrentReadInputStream.getReadInputStream(maxReads, true, ffin1, ffin2, null, null);
jpayne@68: 			cris.start();
jpayne@68: 			if(verbose){outstream.println("Started cris");}
jpayne@68: 		}
jpayne@68: 		
jpayne@68: 		readsProcessed=0;
jpayne@68: 		basesProcessed=0;
jpayne@68: 		
jpayne@68: 		//Process the read stream
jpayne@68: 		processInner(cris, kca);
jpayne@68: 		
jpayne@68: 		if(verbose){outstream.println("Finished; closing streams.");}
jpayne@68: 		
jpayne@68: 		errorState|=ReadStats.writeAll();
jpayne@68: 		errorState|=ReadWrite.closeStreams(cris);
jpayne@68: 		
jpayne@68: 		t.stop();
jpayne@68: 
jpayne@68: 		outstream.println("Made filter:     \t"+kca.toShortString(filterHashes));
jpayne@68: 		outstream.println("Estimated pivots:      \t"+(long)kca.estimateUniqueKmers(filterHashes));
jpayne@68: 		outstream.println("Estimated pivots >1x:  \t"+(long)kca.estimateUniqueKmers(filterHashes, minCount));
jpayne@68: 		
jpayne@68: 		outstream.println(Tools.timeReadsBasesProcessed(t, readsProcessed, basesProcessed, 8));
jpayne@68: 		
jpayne@68: 		if(errorState){
jpayne@68: 			Clumpify.sharedErrorState=true;
jpayne@68: 			throw new RuntimeException(getClass().getName()+" terminated in an error state; the output may be corrupt.");
jpayne@68: 		}
jpayne@68: 		return kca;
jpayne@68: 	}
jpayne@68: 	
jpayne@68: 	/** Manage threads */
jpayne@68: 	public static KCountArray makeKcaStatic(final ConcurrentReadInputStream cris, int k, int minCount, boolean amino){
jpayne@68: 
jpayne@68: 		KmerComparator kc=new KmerComparator(k, false, false);
jpayne@68: 		int cbits=2;
jpayne@68: 		while((1L<<cbits)<=minCount){cbits*=2;}
jpayne@68: 		int filterHashes=2;
jpayne@68: 		float fraction=0.1f;
jpayne@68: 		long cells=getCells(fraction, cbits);
jpayne@68: 		ReadCounter rc=new ReadCounter(k, true, false, false, amino);
jpayne@68: 		KCountArray kca=rc.makeKca(null, null, null, cbits, cells, filterHashes, 0, -1, 1, 1, 1, 1, null, 0);
jpayne@68: 		
jpayne@68: 		if(verbose){System.err.println("Making hash threads.");}
jpayne@68: 		final int threads=Shared.threads();
jpayne@68: 		ArrayList<HashThread> alht=new ArrayList<HashThread>(threads);
jpayne@68: 		for(int i=0; i<threads; i++){alht.add(new HashThread(cris, kc, kca, false));}
jpayne@68: 		
jpayne@68: 		if(verbose){System.err.println("Starting threads.");}
jpayne@68: 		for(HashThread ht : alht){ht.start();}
jpayne@68: 		
jpayne@68: 		if(verbose){System.err.println("Waiting for threads.");}
jpayne@68: 		/* Wait for threads to die */
jpayne@68: 		for(HashThread ht : alht){
jpayne@68: 			
jpayne@68: 			/* Wait for a thread to die */
jpayne@68: 			while(ht.getState()!=Thread.State.TERMINATED){
jpayne@68: 				try {
jpayne@68: 					ht.join();
jpayne@68: 				} catch (InterruptedException e) {
jpayne@68: 					e.printStackTrace();
jpayne@68: 				}
jpayne@68: 			}
jpayne@68: 		}
jpayne@68: 		kca.shutdown();
jpayne@68: 		return kca;
jpayne@68: 	}
jpayne@68: 	
jpayne@68: 	/** Manage threads */
jpayne@68: 	public void processInner(final ConcurrentReadInputStream cris, KCountArray kca){
jpayne@68: 		if(verbose){outstream.println("Making comparator.");}
jpayne@68: 		KmerComparator kc=new KmerComparator(k, false, false);
jpayne@68: 		
jpayne@68: 		if(verbose){outstream.println("Making hash threads.");}
jpayne@68: 		final int threads=Shared.threads();
jpayne@68: 		ArrayList<HashThread> alht=new ArrayList<HashThread>(threads);
jpayne@68: 		for(int i=0; i<threads; i++){alht.add(new HashThread(cris, kc, kca, ecco));}
jpayne@68: 		
jpayne@68: 		if(verbose){outstream.println("Starting threads.");}
jpayne@68: 		for(HashThread ht : alht){ht.start();}
jpayne@68: 		
jpayne@68: 		if(verbose){outstream.println("Waiting for threads.");}
jpayne@68: 		/* Wait for threads to die */
jpayne@68: 		for(HashThread ht : alht){
jpayne@68: 			
jpayne@68: 			/* Wait for a thread to die */
jpayne@68: 			while(ht.getState()!=Thread.State.TERMINATED){
jpayne@68: 				try {
jpayne@68: 					ht.join();
jpayne@68: 				} catch (InterruptedException e) {
jpayne@68: 					e.printStackTrace();
jpayne@68: 				}
jpayne@68: 			}
jpayne@68: 			readsProcessed+=ht.readsProcessedT;
jpayne@68: 			basesProcessed+=ht.basesProcessedT;
jpayne@68: 		}
jpayne@68: 		kca.shutdown();
jpayne@68: 	}
jpayne@68: 	
jpayne@68: 	/*--------------------------------------------------------------*/
jpayne@68: 	/*----------------         Inner Classes        ----------------*/
jpayne@68: 	/*--------------------------------------------------------------*/
jpayne@68: 	
jpayne@68: 	private static class HashThread extends Thread{
jpayne@68: 
jpayne@68: 		HashThread(ConcurrentReadInputStream cris_, KmerComparator kc_, KCountArray kca_, boolean ecco_){
jpayne@68: 			cris=cris_;
jpayne@68: 			kc=kc_;
jpayne@68: 			kca=kca_;
jpayne@68: 			ecco=ecco_;
jpayne@68: 		}
jpayne@68: 
jpayne@68: 		@Override
jpayne@68: 		public void run(){
jpayne@68: 			
jpayne@68: 			ListNum<Read> ln=cris.nextList();
jpayne@68: 			ArrayList<Read> reads=(ln!=null ? ln.list : null);
jpayne@68: 			
jpayne@68: 			while(ln!=null && reads!=null && reads.size()>0){//ln!=null prevents a compiler potential null access warning
jpayne@68: 				for(Read r1 : reads){
jpayne@68: 					Read r2=r1.mate;
jpayne@68: 					readsProcessedT+=r1.pairCount();
jpayne@68: 					basesProcessedT+=r1.pairLength();
jpayne@68: 					if(ecco && r2!=null){
jpayne@68: 						if(r2!=null){BBMerge.findOverlapStrict(r1, r2, true);}
jpayne@68: 					}
jpayne@68: 					{
jpayne@68: 						final long kmer=kc.hash(r1, null, 0, false);
jpayne@68: 						if(kmer>=0){
jpayne@68: 							kca.increment(kmer);
jpayne@68: 						}
jpayne@68: 					}
jpayne@68: 					if(r2!=null){
jpayne@68: 						final long kmer=kc.hash(r2, null, 0, false);
jpayne@68: 						if(kmer>=0){
jpayne@68: 							kca.increment(kmer);
jpayne@68: 						}
jpayne@68: 					}
jpayne@68: 				}
jpayne@68: 				cris.returnList(ln);
jpayne@68: 				ln=cris.nextList();
jpayne@68: 				reads=(ln!=null ? ln.list : null);
jpayne@68: 			}
jpayne@68: 			if(ln!=null){
jpayne@68: 				cris.returnList(ln.id, ln.list==null || ln.list.isEmpty());
jpayne@68: 			}
jpayne@68: 		}
jpayne@68: 		
jpayne@68: 		final ConcurrentReadInputStream cris;
jpayne@68: 		final KmerComparator kc;
jpayne@68: 		final KCountArray kca;
jpayne@68: 		final boolean ecco;
jpayne@68: 		
jpayne@68: 		protected long readsProcessedT=0;
jpayne@68: 		protected long basesProcessedT=0;
jpayne@68: 	}
jpayne@68: 	
jpayne@68: 	/*--------------------------------------------------------------*/
jpayne@68: 	/*----------------         Inner Methods        ----------------*/
jpayne@68: 	/*--------------------------------------------------------------*/
jpayne@68: 	
jpayne@68: 	/*--------------------------------------------------------------*/
jpayne@68: 	/*----------------            Fields            ----------------*/
jpayne@68: 	/*--------------------------------------------------------------*/
jpayne@68: 	
jpayne@68: 	private int k=31;
jpayne@68: 	private int minCount=2;
jpayne@68: 	
jpayne@68: 	/*--------------------------------------------------------------*/
jpayne@68: 	/*----------------          I/O Fields          ----------------*/
jpayne@68: 	/*--------------------------------------------------------------*/
jpayne@68: 	
jpayne@68: 	private String in1=null;
jpayne@68: 	private String in2=null;
jpayne@68: 	
jpayne@68: 	private String extin=null;
jpayne@68: 	
jpayne@68: 	/*--------------------------------------------------------------*/
jpayne@68: 	
jpayne@68: 	protected long readsProcessed=0;
jpayne@68: 	protected long basesProcessed=0;
jpayne@68: 	
jpayne@68: 	private long maxReads=-1;
jpayne@68: 	private boolean ecco=false;
jpayne@68: 	
jpayne@68: 	/*--------------------------------------------------------------*/
jpayne@68: 	/*----------------         Final Fields         ----------------*/
jpayne@68: 	/*--------------------------------------------------------------*/
jpayne@68: 	
jpayne@68: 	private final FileFormat ffin1;
jpayne@68: 	private final FileFormat ffin2;
jpayne@68: 	
jpayne@68: 	/*--------------------------------------------------------------*/
jpayne@68: 	/*----------------        Common Fields         ----------------*/
jpayne@68: 	/*--------------------------------------------------------------*/
jpayne@68: 	
jpayne@68: 	private PrintStream outstream=System.err;
jpayne@68: 	public static boolean verbose=false;
jpayne@68: 	public boolean errorState=false;
jpayne@68: 	
jpayne@68: }