jpayne@68: package aligner;
jpayne@68: 
jpayne@68: import java.io.PrintStream;
jpayne@68: import java.util.ArrayList;
jpayne@68: import java.util.concurrent.atomic.AtomicInteger;
jpayne@68: 
jpayne@68: import fileIO.ByteFile;
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 sketch.SketchObject;
jpayne@68: import stream.ConcurrentReadInputStream;
jpayne@68: import stream.FastaReadInputStream;
jpayne@68: import stream.Read;
jpayne@68: import template.Accumulator;
jpayne@68: import template.ThreadWaiter;
jpayne@68: 
jpayne@68: /**
jpayne@68:  * Aligns all sequences to all sequences and produces an identity matrix.
jpayne@68:  * 
jpayne@68:  * @author Brian Bushnell
jpayne@68:  * @date January 27, 2020
jpayne@68:  *
jpayne@68:  */
jpayne@68: public class AllToAll implements Accumulator<AllToAll.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: 		AllToAll x=new AllToAll(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 AllToAll(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: 		{//Parse the arguments
jpayne@68: 			final Parser parser=parse(args);
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: 			in1=parser.in1;
jpayne@68: 			qfin1=parser.qfin1;
jpayne@68: 			extin=parser.extin;
jpayne@68: 
jpayne@68: 			out1=parser.out1;
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: 		ffout1=FileFormat.testOutput(out1, FileFormat.TXT, null, true, overwrite, append, ordered);
jpayne@68: 
jpayne@68: 		//Create input FileFormat objects
jpayne@68: 		ffin1=FileFormat.testInput(in1, FileFormat.FASTQ, extin, 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: 		parser.out1="stdout.txt";
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("ordered")){
jpayne@68: 				ordered=Parse.parseBoolean(b);
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(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: 		in1=Tools.fixExtension(in1);
jpayne@68: 		qfin1=Tools.fixExtension(qfin1);
jpayne@68: 	}
jpayne@68: 	
jpayne@68: 	/** Ensure files can be read and written */
jpayne@68: 	private void checkFileExistence(){
jpayne@68: 		
jpayne@68: 		//Ensure output files can be written
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 file "+out1+"\n");
jpayne@68: 		}
jpayne@68: 		
jpayne@68: 		//Ensure input files can be read
jpayne@68: 		if(!Tools.testInputFiles(false, true, in1)){
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, in1, out1)){
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: 		//Ensure there is an input file
jpayne@68: 		if(in1==null){throw new RuntimeException("Error - at least one input file is required.");}
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 reads */
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: 		//Reset counters
jpayne@68: 		readsProcessed=alignments=0;
jpayne@68: 		basesProcessed=0;
jpayne@68: 		
jpayne@68: 		//Fetch data
jpayne@68: 		reads=ConcurrentReadInputStream.getReads(maxReads, true, ffin1, null, qfin1, null); //TODO:  Note that this does not return the error state
jpayne@68: 		results=new float[reads.size()][];
jpayne@68: 		
jpayne@68: 		outstream.println("Loaded "+reads.size()+" sequences.");
jpayne@68: 		
jpayne@68: 		//Process the reads in separate threads
jpayne@68: 		spawnThreads();
jpayne@68: 		mirrorMatrix(results);
jpayne@68: 		if(verbose){outstream.println("Finished alignment.");}
jpayne@68: 		
jpayne@68: 		printResults();
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.number("Alignments:", alignments, 8));
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 static void mirrorMatrix(float[][] matrix){
jpayne@68: 		for(int i=0; i<matrix.length; i++) {
jpayne@68: 			for(int j=i; j<matrix.length; j++) {
jpayne@68: 				assert(matrix[i][j]==0) : matrix[i][j];
jpayne@68: 				matrix[i][j]=(i==j ? 1 : matrix[j][i]);
jpayne@68: 			}
jpayne@68: 		}
jpayne@68: 	}
jpayne@68: 	
jpayne@68: 	private void printResults(){
jpayne@68: 		if(ffout1==null){return;}
jpayne@68: 		ByteStreamWriter bsw=new ByteStreamWriter(ffout1);
jpayne@68: 		bsw.start();
jpayne@68: 		final int max=reads.size();
jpayne@68: 		bsw.print("Name");
jpayne@68: 		for(int rnum=0; rnum<max; rnum++){
jpayne@68: 			bsw.tab().print(reads.get(rnum).id);
jpayne@68: 		}
jpayne@68: 		bsw.println();
jpayne@68: 		for(int qnum=0; qnum<max; qnum++){
jpayne@68: 			bsw.print(reads.get(qnum).id);
jpayne@68: 			final float[] scores=results[qnum];
jpayne@68: 			for(int rnum=0; rnum<max; rnum++){
jpayne@68: 				bsw.tab().print(100*scores[rnum], 2);
jpayne@68: 			}
jpayne@68: 			bsw.println();
jpayne@68: 		}
jpayne@68: 		bsw.poisonAndWait();
jpayne@68: 	}
jpayne@68: 	
jpayne@68: 	/*--------------------------------------------------------------*/
jpayne@68: 	/*----------------       Thread Management      ----------------*/
jpayne@68: 	/*--------------------------------------------------------------*/
jpayne@68: 	
jpayne@68: 	/** Spawn process threads */
jpayne@68: 	private void spawnThreads(){
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: 		AtomicInteger atom=new AtomicInteger(0);
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(reads, results, atom, i));
jpayne@68: 		}
jpayne@68: 		
jpayne@68: 		//Start the threads and wait for them to finish
jpayne@68: 		boolean success=ThreadWaiter.startAndWait(alpt, this);
jpayne@68: 		errorState&=!success;
jpayne@68: 		
jpayne@68: 		//Do anything necessary after processing
jpayne@68: 		
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: 		alignments+=pt.alignmentsT;
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: 	/*--------------------------------------------------------------*/
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: 	static class ProcessThread extends Thread {
jpayne@68: 		
jpayne@68: 		//Constructor
jpayne@68: 		ProcessThread(final ArrayList<Read> reads_, float[][] results_, final AtomicInteger atom_, final int tid_){
jpayne@68: 			reads=reads_;
jpayne@68: 			results=results_;
jpayne@68: 			atom=atom_;
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: 			for(int next=atom.getAndIncrement(); next<reads.size(); next=atom.getAndIncrement()){
jpayne@68: 				processQuery(next);
jpayne@68: 			}
jpayne@68: 			
jpayne@68: 		}
jpayne@68: 		
jpayne@68: 		void processQuery(final int qnum){
jpayne@68: 			final Read query=reads.get(qnum);
jpayne@68: 			final float[] scores=new float[reads.size()];
jpayne@68: 			readsProcessedT++;
jpayne@68: 			basesProcessedT+=query.length();
jpayne@68: 			for(int rnum=0; rnum<qnum; rnum++){
jpayne@68: 				final Read ref=reads.get(rnum);
jpayne@68: 				float identity=SketchObject.align(query.bases, ref.bases);
jpayne@68: 				scores[rnum]=identity;
jpayne@68: 				alignmentsT++;
jpayne@68: 			}
jpayne@68: 			synchronized(results){
jpayne@68: 				results[qnum]=scores;
jpayne@68: 			}
jpayne@68: 		}
jpayne@68: 		
jpayne@68: 		/**
jpayne@68: 		 * Process a read or a read pair.
jpayne@68: 		 * @param r1 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: 		boolean processReadPair(final Read r1, final Read r2){
jpayne@68: 			throw new RuntimeException("TODO: Implement this method."); //TODO
jpayne@68: //			return true;
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 alignmentsT=0;
jpayne@68: 		/** Number of bases retained by this thread */
jpayne@68: 		protected long basesOutT=0;
jpayne@68: 		
jpayne@68: 		/** True only if this thread has completed successfully */
jpayne@68: 		boolean success=false;
jpayne@68: 		
jpayne@68: 		/** Thread ID */
jpayne@68: 		final int tid;
jpayne@68: 		
jpayne@68: 		final ArrayList<Read> reads;
jpayne@68: 		final float[][] results;
jpayne@68: 		final AtomicInteger atom;
jpayne@68: 	}
jpayne@68: 	
jpayne@68: 	/*--------------------------------------------------------------*/
jpayne@68: 	/*----------------            Fields            ----------------*/
jpayne@68: 	/*--------------------------------------------------------------*/
jpayne@68: 
jpayne@68: 	/** Primary input file path */
jpayne@68: 	private String in1=null;
jpayne@68: 	
jpayne@68: 	private String qfin1=null;
jpayne@68: 
jpayne@68: 	/** Primary output file path */
jpayne@68: 	private String out1=null;
jpayne@68: 	
jpayne@68: 	/** Override input file extension */
jpayne@68: 	private String extin=null;
jpayne@68: 	
jpayne@68: 	/*--------------------------------------------------------------*/
jpayne@68: 
jpayne@68: 	ArrayList<Read> reads;
jpayne@68: 	float[][] results;
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 alignments=0;
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: 	/*----------------         Final Fields         ----------------*/
jpayne@68: 	/*--------------------------------------------------------------*/
jpayne@68: 
jpayne@68: 	/** Primary input file */
jpayne@68: 	private final FileFormat ffin1;
jpayne@68: 	
jpayne@68: 	/** Primary output file */
jpayne@68: 	private final FileFormat ffout1;
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: }