jpayne@68: package bloom;
jpayne@68: 
jpayne@68: import java.util.ArrayList;
jpayne@68: import java.util.Arrays;
jpayne@68: 
jpayne@68: import dna.AminoAcid;
jpayne@68: import fileIO.ReadWrite;
jpayne@68: import shared.Timer;
jpayne@68: import stream.ConcurrentGenericReadInputStream;
jpayne@68: import stream.FastqReadInputStream;
jpayne@68: import stream.Read;
jpayne@68: import structures.ListNum;
jpayne@68: 
jpayne@68: /**
jpayne@68:  * @author Brian Bushnell
jpayne@68:  * @date Jul 5, 2012
jpayne@68:  *
jpayne@68:  */
jpayne@68: public class TestLargeKmer {
jpayne@68: 	
jpayne@68: 	public static void main(String args[]){
jpayne@68: 		Timer t=new Timer();
jpayne@68: 		
jpayne@68: 		String fname1=args[0];
jpayne@68: 		String fname2=(args.length>4 || args[1].contains(".") ? args[1] : null);
jpayne@68: 		int k=Integer.parseInt(args[args.length-3]);
jpayne@68: 		int cbits=Integer.parseInt(args[args.length-2]);
jpayne@68: 		int k2=Integer.parseInt(args[args.length-1]);
jpayne@68: 		
jpayne@68: 		KCountArray2 counts=KmerCount3.countFastq(fname1, fname2, k, cbits);
jpayne@68: 		long[] counts2=countK2(fname1, fname2, k, counts, k2);
jpayne@68: 		
jpayne@68: 		t.stop();
jpayne@68: 		System.out.println("Finished counting; time = "+t+"\n");
jpayne@68: 		
jpayne@68: 		for(int i=0; i<counts2.length; i++){
jpayne@68: 			System.out.println(i+":\t"+counts2[i]);
jpayne@68: 		}
jpayne@68: 	}
jpayne@68: 	
jpayne@68: 	public static long[] countK2(String fname1, String fname2, int k, int cbits, int k2){
jpayne@68: 		KCountArray2 counts=KmerCount3.countFastq(fname1, fname2, k, cbits);
jpayne@68: 		return countK2(fname1, fname2, k, counts, k2);
jpayne@68: 	}
jpayne@68: 	
jpayne@68: 	public static long[] countK2(String fname1, String fname2, int k, KCountArray2 counts1, int k2){
jpayne@68: 		assert(k>=1 && k<20);
jpayne@68: 		final int kbits=2*k;
jpayne@68: 		final long mask=(kbits>63 ? -1L : ~((-1L)<<kbits));
jpayne@68: 		FastqReadInputStream fris1=new FastqReadInputStream(fname1, false);
jpayne@68: 		FastqReadInputStream fris2=(fname2==null ? null : new FastqReadInputStream(fname2, false));
jpayne@68: 		ConcurrentGenericReadInputStream cris=new ConcurrentGenericReadInputStream(fris1, fris2, KmerCount3.maxReads);
jpayne@68: 		
jpayne@68: 		cris.start();
jpayne@68: 		System.err.println("Started cris");
jpayne@68: 		boolean paired=cris.paired();
jpayne@68: 		
jpayne@68: 		long kmer=0; //current kmer
jpayne@68: 		int len=0;  //distance since last contig start or ambiguous base
jpayne@68: 		
jpayne@68: 		
jpayne@68: 		final long[] upperBound=new long[BOUND_LEN]; //Lowest upper bound provable of kmer count
jpayne@68: 		final int[] ring=new int[k2-k+1];
jpayne@68: 		final int[] subcount=new int[BOUND_LEN];
jpayne@68: 		final int maxValue=subcount.length-1;
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: 			if(reads!=null && !reads.isEmpty()){
jpayne@68: 				Read r=reads.get(0);
jpayne@68: 				assert(paired==(r.mate!=null));
jpayne@68: 			}
jpayne@68: 			
jpayne@68: 			while(ln!=null && reads!=null && reads.size()>0){//ln!=null prevents a compiler potential null access warning
jpayne@68: 				//System.err.println("reads.size()="+reads.size());
jpayne@68: 				for(Read r : reads){
jpayne@68: 					
jpayne@68: 					len=0;
jpayne@68: 					kmer=0;
jpayne@68: 					Arrays.fill(subcount, 0);
jpayne@68: 					
jpayne@68: 					byte[] bases=r.bases;
jpayne@68: 					byte[] quals=r.quality;
jpayne@68: 					for(int i=0; i<bases.length; i++){
jpayne@68: 						byte b=bases[i];
jpayne@68: 						int x=AminoAcid.baseToNumber[b];
jpayne@68: 						
jpayne@68: 						int ringpos=i%ring.length;
jpayne@68: 						int old=ring[ringpos];
jpayne@68: 						int value=0;
jpayne@68: 						
jpayne@68: 						if(x<0 || quals[i]<KmerCount3.minQuality){
jpayne@68: 							len=0;
jpayne@68: 							kmer=0;
jpayne@68: 						}else{
jpayne@68: 							kmer=((kmer<<2)|x)&mask;
jpayne@68: 							len++;
jpayne@68: 							
jpayne@68: 							if(len>=k){
jpayne@68: 								value=counts1.read(kmer);
jpayne@68: 							}
jpayne@68: 						}
jpayne@68: 						value=min(value, maxValue);
jpayne@68: 						
jpayne@68: 						ring[ringpos]=value;
jpayne@68: 						subcount[value]++;
jpayne@68: 						
jpayne@68: 						if(i>=ring.length){
jpayne@68: 							subcount[old]--;
jpayne@68: 						}
jpayne@68: 						
jpayne@68: 						if(len>=k2){
jpayne@68: 							int sub=0;
jpayne@68: 							while(sub<subcount.length && subcount[sub]==0){sub++;}
jpayne@68: 							assert(sub<subcount.length);
jpayne@68: 							upperBound[sub]++;
jpayne@68: 						}
jpayne@68: 						
jpayne@68: 					}
jpayne@68: 					
jpayne@68: 					if(r.mate!=null){
jpayne@68: 						bases=r.mate.bases;
jpayne@68: 						quals=r.mate.quality;
jpayne@68: 
jpayne@68: 						len=0;
jpayne@68: 						kmer=0;
jpayne@68: 						Arrays.fill(subcount, 0);
jpayne@68: 						for(int i=0; i<bases.length; i++){
jpayne@68: 							byte b=bases[i];
jpayne@68: 							int x=AminoAcid.baseToNumber[b];
jpayne@68: 							
jpayne@68: 							int ringpos=i%ring.length;
jpayne@68: 							int old=ring[ringpos];
jpayne@68: 							int value=0;
jpayne@68: 							
jpayne@68: 							if(x<0 || quals[i]<KmerCount3.minQuality){
jpayne@68: 								len=0;
jpayne@68: 								kmer=0;
jpayne@68: 							}else{
jpayne@68: 								kmer=((kmer<<2)|x)&mask;
jpayne@68: 								len++;
jpayne@68: 								
jpayne@68: 								if(len>=k){
jpayne@68: 									value=counts1.read(kmer);
jpayne@68: 								}
jpayne@68: 							}
jpayne@68: 							value=min(value, maxValue);
jpayne@68: 							
jpayne@68: 							ring[ringpos]=value;
jpayne@68: 							subcount[value]++;
jpayne@68: 							
jpayne@68: 							if(i>=ring.length){
jpayne@68: 								subcount[old]--;
jpayne@68: 							}
jpayne@68: 							
jpayne@68: 							if(len>=k2){
jpayne@68: 								int sub=0;
jpayne@68: 								while(sub<subcount.length && subcount[sub]==0){sub++;}
jpayne@68: 								assert(sub<subcount.length);
jpayne@68: 								upperBound[sub]++;
jpayne@68: 							}
jpayne@68: 							
jpayne@68: 						}
jpayne@68: 					}
jpayne@68: 					
jpayne@68: 				}
jpayne@68: 				//System.err.println("returning list");
jpayne@68: 				cris.returnList(ln);
jpayne@68: 				//System.err.println("fetching list");
jpayne@68: 				ln=cris.nextList();
jpayne@68: 				reads=(ln!=null ? ln.list : null);
jpayne@68: 			}
jpayne@68: 			System.err.println("Finished reading");
jpayne@68: 			cris.returnList(ln);
jpayne@68: 			System.err.println("Returned list");
jpayne@68: 			ReadWrite.closeStreams(cris);
jpayne@68: 			System.err.println("Closed stream");
jpayne@68: 		}
jpayne@68: 		
jpayne@68: 		return upperBound;
jpayne@68: 	}
jpayne@68: 	
jpayne@68: 	public static final int min(int x, int y){return x<y ? x : y;}
jpayne@68: 	public static final int max(int x, int y){return x>y ? x : y;}
jpayne@68: 	
jpayne@68: 	public static final int BOUND_LEN=256;
jpayne@68: 	
jpayne@68: }