jpayne@68: package icecream;
jpayne@68: 
jpayne@68: import java.util.ArrayList;
jpayne@68: 
jpayne@68: import shared.Tools;
jpayne@68: import stream.Read;
jpayne@68: import stream.SamLine;
jpayne@68: import structures.IntList;
jpayne@68: 
jpayne@68: /**
jpayne@68:  * Container for the list of reads from a single
jpayne@68:  * PacBio ZMW.
jpayne@68:  * @author Brian Bushnell
jpayne@68:  * @date June 5, 2020
jpayne@68:  */
jpayne@68: public class ZMW extends ArrayList<Read> {
jpayne@68: 	
jpayne@68: 	/**
jpayne@68: 	 * For serialization.
jpayne@68: 	 */
jpayne@68: 	private static final long serialVersionUID = -2580124131008824113L;
jpayne@68: 
jpayne@68: 	public ZMW(){super();}
jpayne@68: 	
jpayne@68: 	public ZMW(int initialSize){super(initialSize);}
jpayne@68: 
jpayne@68: 	public long countBases(){
jpayne@68: 		long x=0;
jpayne@68: 		for(Read r : this){
jpayne@68: 			x+=r.length();
jpayne@68: 		}
jpayne@68: 		return x;
jpayne@68: 	}
jpayne@68: 	
jpayne@68: 	public int medianLength(boolean includeDiscarded){
jpayne@68: 		if(size()<3){return -1;}
jpayne@68: 		IntList lengths=new IntList(size()-2);
jpayne@68: 		
jpayne@68: 		for(int i=1; i<size()-1; i++){
jpayne@68: 			Read r=get(i);
jpayne@68: 			if(includeDiscarded || !r.discarded()){
jpayne@68: 				lengths.add(get(i).length());
jpayne@68: 			}
jpayne@68: 		}
jpayne@68: 		lengths.sort();
jpayne@68: 		int median=lengths.get(lengths.size/2);
jpayne@68: 		return median;
jpayne@68: 	}
jpayne@68: 	
jpayne@68: 	public int longestLength(boolean includeDiscarded){
jpayne@68: 		int max=0;
jpayne@68: 		for(Read r : this){
jpayne@68: 			if(includeDiscarded || !r.discarded()){
jpayne@68: 				max=Tools.max(max, r.length());
jpayne@68: 			}
jpayne@68: 		}
jpayne@68: 		return max;
jpayne@68: 	}
jpayne@68: 	
jpayne@68: 	public Read medianRead(boolean includeDiscarded){
jpayne@68: 		int len=medianLength(includeDiscarded);
jpayne@68: 		if(len<0){return longestRead(includeDiscarded);}
jpayne@68: 		for(int i=1; i<size()-1; i++){
jpayne@68: 			Read r=get(i);
jpayne@68: 			if((includeDiscarded || !r.discarded()) && r.length()==len){
jpayne@68: 				return r;
jpayne@68: 			}
jpayne@68: 		}
jpayne@68: 		return null;
jpayne@68: 	}
jpayne@68: 	
jpayne@68: 	public Read longestRead(boolean includeDiscarded){
jpayne@68: 		Read max=null;
jpayne@68: 		for(Read r : this){
jpayne@68: 			if((includeDiscarded || !r.discarded()) && (max==null || r.length()>max.length())){max=r;}
jpayne@68: 		}
jpayne@68: 		return max;
jpayne@68: 	}
jpayne@68: 	
jpayne@68: 	public int zid(){
jpayne@68: 		if(zid==-1){parseZID();}
jpayne@68: 		return zid;
jpayne@68: 	}
jpayne@68: 	
jpayne@68: 	private int parseZID(){
jpayne@68: 		return (size()<1 ? -1 : PBHeader.parseZMW(get(0).id));
jpayne@68: 	}
jpayne@68: 	
jpayne@68: 	public static void fixReadHeader(Read r, int leftTrim, int rightTrim){
jpayne@68: 		leftTrim=Tools.max(0, leftTrim);
jpayne@68: 		rightTrim=Tools.max(0, rightTrim);
jpayne@68: 		if(leftTrim<1 && rightTrim<1){return;}
jpayne@68: 		final int idx=r.id.lastIndexOf('/');
jpayne@68: 		if(idx>0 && idx<r.id.length()-3){
jpayne@68: 			String prefix=r.id.substring(0, idx+1);
jpayne@68: 			String suffix=r.id.substring(idx+1);
jpayne@68: 			if(suffix.indexOf('_')>0){
jpayne@68: 				String coords=suffix, comment="";
jpayne@68: 				int tab=suffix.indexOf('\t');
jpayne@68: 				if(tab<0){tab=suffix.indexOf(' ');}
jpayne@68: 				if(tab>0){
jpayne@68: 					coords=coords.substring(0, tab);
jpayne@68: 					comment=coords.substring(tab);
jpayne@68: 				}
jpayne@68: 				String[] split=Tools.underscorePattern.split(coords);
jpayne@68: 				int left=Integer.parseInt(split[0]);
jpayne@68: 				int right=Integer.parseInt(split[1]);
jpayne@68: 				left+=leftTrim;
jpayne@68: 				right-=rightTrim;
jpayne@68: 				if(left>right){left=right;}
jpayne@68: 				
jpayne@68: 				if(right-left!=r.length()){right=left+r.length();}
jpayne@68: //				System.err.println(r.length()+", "+(right-left));
jpayne@68: 				
jpayne@68: 				r.id=prefix+left+"_"+right+comment;
jpayne@68: 				final SamLine sl=r.samline;
jpayne@68: 				if(sl!=null){
jpayne@68: 					sl.qname=r.id;
jpayne@68: 					if(sl.optional!=null){
jpayne@68: 						for(int i=0; i<sl.optional.size(); i++){
jpayne@68: 							String s=sl.optional.get(i);
jpayne@68: 							if(s.startsWith("qe:i:")){
jpayne@68: 								s="qe:i:"+right;
jpayne@68: 								sl.optional.set(i, s);
jpayne@68: 							}else if(s.startsWith("qs:i:")){
jpayne@68: 								s="qs:i:"+left;
jpayne@68: 								sl.optional.set(i, s);
jpayne@68: 							}
jpayne@68: 						}
jpayne@68: 					}
jpayne@68: 				}
jpayne@68: 			}
jpayne@68: 		}
jpayne@68: 	}
jpayne@68: 	
jpayne@68: 	public void setDiscarded(boolean b){
jpayne@68: 		for(Read r : this){
jpayne@68: 			r.setDiscarded(b);
jpayne@68: 		}
jpayne@68: 	}
jpayne@68: 
jpayne@68: 	public int[] lengths() {
jpayne@68: 		final int size=size();
jpayne@68: 		int[] array=new int[size];
jpayne@68: 		for(int i=0; i<size; i++){
jpayne@68: 			Read r=get(i);
jpayne@68: 			array[i]=r==null ? -1 : r.length();
jpayne@68: 		}
jpayne@68: 		return array;
jpayne@68: 	}
jpayne@68: 	
jpayne@68: 	public float estimatePasses(){
jpayne@68: 		final int size=size();
jpayne@68: 		if(size<1){return 0;}
jpayne@68: 		else if(size==1){return 0.25f;}
jpayne@68: 		else if(size==2){return 0.5f;}
jpayne@68: 		
jpayne@68: 		int median=medianLength(true);
jpayne@68: 		int first=first().length();
jpayne@68: 		int last=last().length();
jpayne@68: 
jpayne@68: 		return size-2+estimatePasses(first, median)+estimatePasses(last, median);
jpayne@68: 	}
jpayne@68: 	
jpayne@68: 	private float estimatePasses(int len, int median){
jpayne@68: 		float ratio=len/(float)median;
jpayne@68: 		//TODO: I want this to be more asymptotic
jpayne@68: 		return Tools.min(0.99f, ratio/(1+0.05f*ratio));
jpayne@68: 	}
jpayne@68: 
jpayne@68: 	public boolean discarded() {
jpayne@68: 		for(Read r : this){
jpayne@68: 			if(!r.discarded()){return false;}
jpayne@68: 		}
jpayne@68: 		return true;
jpayne@68: 	}
jpayne@68: 	
jpayne@68: 	/** 
jpayne@68: 	 * Identifier assigned by streamer, not by PacBio.
jpayne@68: 	 * First identifier is 0, then 1, etc.
jpayne@68: 	 */
jpayne@68: 	public long id;
jpayne@68: 	
jpayne@68: 	/** 
jpayne@68: 	 * ZMW ID assigned by PacBio.
jpayne@68: 	 */
jpayne@68: 	private int zid=-1;
jpayne@68: 
jpayne@68: 	public Read first(){return get(0);}
jpayne@68: 	public Read last(){return get(size()-1);}
jpayne@68: 	
jpayne@68: }