jpayne@68: package sketch;
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 dna.AminoAcid;
jpayne@68: import fileIO.ByteFile;
jpayne@68: import fileIO.ByteFile1;
jpayne@68: import fileIO.ByteFile2;
jpayne@68: import fileIO.ByteStreamWriter;
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.ConcurrentGenericReadInputStream;
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: import structures.LongHashSet;
jpayne@68: 
jpayne@68: /**
jpayne@68:  * @author Brian Bushnell
jpayne@68:  * @date Oct 17, 2014
jpayne@68:  *
jpayne@68:  */
jpayne@68: public class InvertKey extends SketchObject {
jpayne@68: 	
jpayne@68: 	public static void main(String[] args){
jpayne@68: 		Timer t=new Timer();
jpayne@68: 		InvertKey x=new InvertKey(args);
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: 	public InvertKey(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: 		Shared.capBuffers(4);
jpayne@68: 		ReadWrite.USE_PIGZ=ReadWrite.USE_UNPIGZ=true;
jpayne@68: 		ReadWrite.MAX_ZIP_THREADS=Shared.threads();
jpayne@68: 		
jpayne@68: 		FASTQ.TEST_INTERLEAVED=FASTQ.FORCE_INTERLEAVED=false;
jpayne@68: 		int k_=32, k2_=0;
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(a.equals("verbose")){
jpayne@68: 				verbose=Parse.parseBoolean(b);
jpayne@68: 				ByteFile1.verbose=verbose;
jpayne@68: 				ByteFile2.verbose=verbose;
jpayne@68: 				stream.FastaReadInputStream.verbose=verbose;
jpayne@68: 				ConcurrentGenericReadInputStream.verbose=verbose;
jpayne@68: 				stream.FastqReadInputStream.verbose=verbose;
jpayne@68: 				ReadWrite.verbose=verbose;
jpayne@68: 			}else if(a.equals("key")){
jpayne@68: 				keyString=b;
jpayne@68: 			}else if(a.equals("out")){
jpayne@68: 				out1=b;
jpayne@68: 			}else if(a.equalsIgnoreCase("k")){
jpayne@68: 				assert(b!=null) : "Bad parameter: "+arg;
jpayne@68: 				if(b.indexOf(',')>=0){
jpayne@68: 					String[] bsplit=b.split(",");
jpayne@68: 					assert(bsplit.length==2) : "Bad argument "+arg;
jpayne@68: 					int x=Integer.parseInt(bsplit[0]);
jpayne@68: 					int y=Integer.parseInt(bsplit[1]);
jpayne@68: 					k_=Tools.max(x, y);
jpayne@68: 					k2_=Tools.min(x, y);
jpayne@68: 					if(k_==k2_){k2_=0;}
jpayne@68: 				}else{
jpayne@68: 					k_=Integer.parseInt(b);
jpayne@68: 					k2_=0;
jpayne@68: 				}
jpayne@68: 			}else if(a.equalsIgnoreCase("printonce")){
jpayne@68: 				printOnce=Parse.parseBoolean(b);
jpayne@68: 			}else if(parser.in1==null && i==0 && !arg.contains("=") && (arg.toLowerCase().startsWith("stdin") || new File(arg).exists())){
jpayne@68: 				parser.in1=arg;
jpayne@68: 			}else if(parser.out1==null && i==1 && !arg.contains("=")){
jpayne@68: 				out1=arg;
jpayne@68: 			}else if(parser.parse(arg, a, b)){
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: 				//				throw new RuntimeException("Unknown parameter "+args[i]);
jpayne@68: 			}
jpayne@68: 		}
jpayne@68: 		
jpayne@68: 		k=k_;
jpayne@68: 		k2=k2_;
jpayne@68: 		shift=2*k;
jpayne@68: 		shift2=shift-2;
jpayne@68: 		mask=(shift>63 ? -1L : ~((-1L)<<shift)); //Conditional allows K=32
jpayne@68: 		
jpayne@68: 		{//Process parser fields
jpayne@68: 			Parser.processQuality();
jpayne@68: 			
jpayne@68: 			maxReads=parser.maxReads;
jpayne@68: 			
jpayne@68: 			overwrite=ReadStats.overwrite=parser.overwrite;
jpayne@68: 			append=ReadStats.append=parser.append;
jpayne@68: 			
jpayne@68: 			in1=parser.in1;
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){
jpayne@68: 			ByteFile.FORCE_MODE_BF2=false;
jpayne@68: 			ByteFile.FORCE_MODE_BF1=true;
jpayne@68: 		}
jpayne@68: 
jpayne@68: 		if(out1!=null && out1.equalsIgnoreCase("null")){out1=null;}
jpayne@68: 		
jpayne@68: 		if(!Tools.testOutputFiles(overwrite, append, false, out1)){
jpayne@68: 			outstream.println((out1==null)+", "+out1);
jpayne@68: 			throw new RuntimeException("\n\noverwrite="+overwrite+"; Can't write to output files "+out1+"\n");
jpayne@68: 		}
jpayne@68: 		
jpayne@68: 		ffout1=FileFormat.testOutput(out1, FileFormat.FASTA, null, true, overwrite, append, false);
jpayne@68: 
jpayne@68: 		ffin1=FileFormat.testInput(in1, FileFormat.FASTA, null, true, true);
jpayne@68: 		
jpayne@68: 		SketchObject.postParse();
jpayne@68: 		
jpayne@68: 		if(keyString.indexOf(',')>0){
jpayne@68: 			String[] split=keyString.split(",");
jpayne@68: 			set=new LongHashSet(split.length*2);
jpayne@68: 			for(String s : split){
jpayne@68: 				long x=Long.MAX_VALUE-Sketch.parseA48(s);
jpayne@68: 				set.add(x);
jpayne@68: //				assert(set.contains(x)) : x+", "+set.size()+", "+set.toStringListView();
jpayne@68: 			}
jpayne@68: 			key0=-1;
jpayne@68: //			System.err.println(set.toStringListView()+", "+set.size());
jpayne@68: 			assert(!set.isEmpty());
jpayne@68: 		}else if(keyString.endsWith(".sketch")){
jpayne@68: 			SketchTool tool=new SketchTool(10000, 0, false, false);
jpayne@68: 			Sketch sk=tool.loadSketchesFromFile(keyString, null, 0, 1000000, SketchObject.ONE_SKETCH, 1f, 0f, 0f, (byte)0, false).get(0);
jpayne@68: 			set=new LongHashSet(sk.length()*2);
jpayne@68: 			for(long x : sk.keys){set.add(Long.MAX_VALUE-x);}
jpayne@68: 			key0=-1;
jpayne@68: //			System.err.println(set.toStringListView()+", "+set.size());
jpayne@68: 			assert(!set.isEmpty());
jpayne@68: 		}else{
jpayne@68: 			key0=Long.MAX_VALUE-Sketch.parseA48(keyString);
jpayne@68: 			set=null;
jpayne@68: //			System.err.println(key0);
jpayne@68: 		}
jpayne@68: 	}
jpayne@68: 	
jpayne@68: 	void process(Timer t){
jpayne@68: 		
jpayne@68: 		final ConcurrentReadInputStream cris;
jpayne@68: 		{
jpayne@68: 			cris=ConcurrentReadInputStream.getReadInputStream(maxReads, true, ffin1, null, null, null);
jpayne@68: 			cris.start();
jpayne@68: 			if(verbose){outstream.println("Started cris");}
jpayne@68: 		}
jpayne@68: 		boolean paired=cris.paired();
jpayne@68: //		if(verbose){
jpayne@68: 			if(!ffin1.samOrBam()){outstream.println("Input is being processed as "+(paired ? "paired" : "unpaired"));}
jpayne@68: //		}
jpayne@68: 
jpayne@68: 		final ByteStreamWriter bsw;
jpayne@68: 		if(out1!=null){
jpayne@68: 			fasta=ffout1.fasta() && !out1.endsWith(".txt");
jpayne@68: 			bsw=new ByteStreamWriter(ffout1);
jpayne@68: 			bsw.start();
jpayne@68: 		}else{bsw=null;}
jpayne@68: 		
jpayne@68: 		long readsProcessed=0;
jpayne@68: 		long basesProcessed=0;
jpayne@68: 		boolean finished=false;
jpayne@68: 		
jpayne@68: 		{
jpayne@68: 			
jpayne@68: 			ListNum<Read> ln=cris.nextList();
jpayne@68: 			ArrayList<Read> reads=(ln!=null ? ln.list : null);
jpayne@68: 			
jpayne@68: //			outstream.println("Fetched "+reads);
jpayne@68: 			
jpayne@68: 			if(reads!=null && !reads.isEmpty()){
jpayne@68: 				Read r=reads.get(0);
jpayne@68: 				assert((ffin1==null || ffin1.samOrBam()) || (r.mate!=null)==cris.paired());
jpayne@68: 			}
jpayne@68: 			
jpayne@68: 			while(reads!=null && reads.size()>0 && !finished){
jpayne@68: 				
jpayne@68: 				for(int idx=0; idx<reads.size() && !finished; idx++){
jpayne@68: 					final Read r1=reads.get(idx);
jpayne@68: 
jpayne@68: 					finished=invert(key0, r1, bsw);
jpayne@68: 					
jpayne@68: 					final int initialLength1=r1.length();
jpayne@68: 					
jpayne@68: 					readsProcessed++;
jpayne@68: 					basesProcessed+=initialLength1;
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: 		errorState|=(ReadWrite.closeStream(cris));
jpayne@68: 		if(bsw!=null){errorState|=bsw.poisonAndWait();}
jpayne@68: 		
jpayne@68: 		t.stop();
jpayne@68: 		outstream.println(Tools.timeReadsBasesProcessed(t, readsProcessed, basesProcessed, 8));
jpayne@68: 		
jpayne@68: 		if(errorState && !finished && maxReads<1){
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 boolean invert(long key2, Read r, ByteStreamWriter bsw) {
jpayne@68: 		final byte[] bases=r.bases;
jpayne@68: 		
jpayne@68: 		long kmer=0;
jpayne@68: 		long rkmer=0;
jpayne@68: 		int len=0;
jpayne@68: 		
jpayne@68: 
jpayne@68: //		System.err.println("Looking for "+key+"\t"+Sketch.toA48(key)+"\t"+Sketch.toA48(Long.MAX_VALUE-key));
jpayne@68: 		
jpayne@68: 		for(int i=0; i<bases.length; i++){
jpayne@68: 			byte b=bases[i];
jpayne@68: 			long x=AminoAcid.baseToNumber[b];
jpayne@68: 			long x2=AminoAcid.baseToComplementNumber[b];
jpayne@68: 			kmer=((kmer<<2)|x)&mask;
jpayne@68: 			rkmer=((rkmer>>>2)|(x2<<shift2))&mask;
jpayne@68: 			if(x<0){len=0; rkmer=0;}else{len++;}
jpayne@68: 			if(len>=k){
jpayne@68: 				kmersProcessed++;
jpayne@68: 				final long hashcode=hash(kmer, rkmer);
jpayne@68: 				boolean found=(key0>=0 ? hashcode==key0 : set.contains(hashcode));
jpayne@68: 				if(found){
jpayne@68: 					if(fasta){bsw.println(">"+Sketch.toA48(Long.MAX_VALUE-hashcode)+" "+(i-k+1)+" "+r.id);}
jpayne@68: 					bsw.println(AminoAcid.kmerToString(Tools.min(kmer, rkmer), k));
jpayne@68: 					if(printOnce){
jpayne@68: 						if(key0>=0){return true;}
jpayne@68: 						else{
jpayne@68: 							set.remove(hashcode);
jpayne@68: 							return set.isEmpty();
jpayne@68: 						}
jpayne@68: 					}
jpayne@68: 				}
jpayne@68: 			}
jpayne@68: 		}
jpayne@68: 		return false;
jpayne@68: 	}
jpayne@68: 	
jpayne@68: 	/*--------------------------------------------------------------*/
jpayne@68: 	
jpayne@68: 	final long key0;
jpayne@68: 	final LongHashSet set;
jpayne@68: 	
jpayne@68: 	final int shift;
jpayne@68: 	final int shift2;
jpayne@68: 	final long mask;
jpayne@68: 	
jpayne@68: 	boolean printOnce=true;
jpayne@68: 	long kmersProcessed=0;
jpayne@68: 	
jpayne@68: 	private String in1=null;
jpayne@68: 	boolean fasta;
jpayne@68: 	boolean sketch;
jpayne@68: 	private String keyString=null;
jpayne@68: 
jpayne@68: 	private String out1="stdout.fa";
jpayne@68: 	
jpayne@68: 	/*--------------------------------------------------------------*/
jpayne@68: 
jpayne@68: 	private long maxReads=-1;
jpayne@68: 	
jpayne@68: 	/*--------------------------------------------------------------*/
jpayne@68: 	
jpayne@68: 	private final FileFormat ffin1;
jpayne@68: 
jpayne@68: 	private final FileFormat ffout1;
jpayne@68: 	
jpayne@68: 	
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: 	private boolean overwrite=false;
jpayne@68: 	private boolean append=false;
jpayne@68: 	
jpayne@68: }