jpayne@68: package gff;
jpayne@68: 
jpayne@68: import java.io.PrintStream;
jpayne@68: 
jpayne@68: import fileIO.ByteFile;
jpayne@68: import fileIO.ByteStreamWriter;
jpayne@68: import fileIO.FileFormat;
jpayne@68: import shared.Parser;
jpayne@68: import shared.PreParser;
jpayne@68: import shared.ReadStats;
jpayne@68: import shared.Timer;
jpayne@68: import shared.Tools;
jpayne@68: import structures.ByteBuilder;
jpayne@68: import var2.VCFLine;
jpayne@68: 
jpayne@68: /**
jpayne@68:  * Stripped out of GffLine into independent class.
jpayne@68:  * @author Brian Bushnell
jpayne@68:  * @date Sep 12, 2018
jpayne@68:  *
jpayne@68:  */
jpayne@68: public class VcfToGff {
jpayne@68: 
jpayne@68: 	/** Translates VCF to GFF */
jpayne@68: 	public static void main(String[] args){
jpayne@68: 		Timer t=new Timer();
jpayne@68: 		PrintStream outstream=System.err;
jpayne@68: 		{//Preparse block for help, config files, and outstream
jpayne@68: 			PreParser pp=new PreParser(args, new Object() { }.getClass().getEnclosingClass(), false);
jpayne@68: 			args=pp.args;
jpayne@68: 			outstream=pp.outstream;
jpayne@68: 			t.outstream=outstream;
jpayne@68: 		}
jpayne@68: 		
jpayne@68: 		Parser parser=new Parser();
jpayne@68: 		String in=null;
jpayne@68: 		String out=null;
jpayne@68: 		boolean overwrite=true, append=false;
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: 			
jpayne@68: 			if(a.equals("in") || a.equals("vcf")){
jpayne@68: 				in=b;
jpayne@68: 			}else if(a.equals("out") || a.equals("gff")){
jpayne@68: 				out=b;
jpayne@68: 			}else if(parser.parse(arg, a, b)){
jpayne@68: 				//do nothing
jpayne@68: 			}else if(in==null && b==null && i==0 && Tools.canRead(arg)){
jpayne@68: 				in=arg;
jpayne@68: 			}else if(in==null && b==null && i==1){
jpayne@68: 				out=arg;
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: 		{//Process parser fields
jpayne@68: 			overwrite=ReadStats.overwrite=parser.overwrite;
jpayne@68: 			append=ReadStats.append=parser.append;
jpayne@68: 		}
jpayne@68: 		
jpayne@68: 		//Ensure output files can be written
jpayne@68: 		if(!Tools.testOutputFiles(overwrite, append, false, out)){
jpayne@68: 			outstream.println((out==null)+", "+out);
jpayne@68: 			throw new RuntimeException("\n\noverwrite="+overwrite+"; Can't write to output files "+out+"\n");
jpayne@68: 		}
jpayne@68: 
jpayne@68: 		//Ensure input files can be read
jpayne@68: 		if(!Tools.testInputFiles(false, true, in)){
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, in, out)){
jpayne@68: 			throw new RuntimeException("\nSome file names were specified multiple times.\n");
jpayne@68: 		}
jpayne@68: 		
jpayne@68: 		translate(in, out, overwrite, append);
jpayne@68: 		t.stop("Time: \t");
jpayne@68: 	}
jpayne@68: 	
jpayne@68: 	/** Translates VCF to GFF */
jpayne@68: 	private static void translate(String in, String out, boolean overwrite, boolean append){
jpayne@68: 		//Create output FileFormat objects
jpayne@68: 		FileFormat ffout=FileFormat.testOutput(out, FileFormat.GFF, "gff", true, overwrite, append, false);
jpayne@68: 
jpayne@68: 		//Create input FileFormat objects
jpayne@68: 		FileFormat ffin=FileFormat.testInput(in, FileFormat.VCF, "vcf", true, true);
jpayne@68: 		
jpayne@68: 		ByteFile bf=ByteFile.makeByteFile(ffin);
jpayne@68: 		ByteStreamWriter bsw=null;
jpayne@68: 		if(ffout!=null){
jpayne@68: 			bsw=new ByteStreamWriter(ffout);
jpayne@68: 			bsw.start();
jpayne@68: 		}
jpayne@68: 		
jpayne@68: 		ByteBuilder bb=new ByteBuilder(17000);
jpayne@68: 		bb.append("##gff-version 3\n");
jpayne@68: 		String header="#seqid	source	type	start	end	score	strand	phase	attributes";
jpayne@68: 		for(byte[] line=bf.nextLine(); line!=null; line=bf.nextLine()){
jpayne@68: 			if(line.length>1){
jpayne@68: 				if(line[0]=='#'){
jpayne@68: 					if(Tools.startsWith(line, "##fileformat") || Tools.startsWith(line, "##FORMAT") || 
jpayne@68: 							Tools.startsWith(line, "##INFO") || Tools.startsWith(line, "#CHROM	POS")){
jpayne@68: 						//skip
jpayne@68: 					}else{
jpayne@68: 						int i=1;
jpayne@68: 						while(i<line.length && line[i]=='#'){i++;}
jpayne@68: 						i--;
jpayne@68: 						bb.append(line, i, line.length-i);
jpayne@68: 						bb.nl();
jpayne@68: 					}
jpayne@68: 				}else{
jpayne@68: 					if(header!=null){
jpayne@68: 						bb.append(header).append('\n');
jpayne@68: 						header=null;
jpayne@68: 					}
jpayne@68: 					VCFLine vline=new VCFLine(line);
jpayne@68: 					GffLine gline=new GffLine(vline);
jpayne@68: 					gline.appendTo(bb);
jpayne@68: 					bb.nl();
jpayne@68: 				}
jpayne@68: 			}
jpayne@68: 			if(bb.length()>=16384){
jpayne@68: 				if(bsw!=null){
jpayne@68: 					bsw.print(bb);
jpayne@68: 				}
jpayne@68: 				bb.clear();
jpayne@68: 			}
jpayne@68: 		}
jpayne@68: 		if(bb.length()>0){
jpayne@68: 			if(bsw!=null){
jpayne@68: 				bsw.print(bb);
jpayne@68: 			}
jpayne@68: 			bb.clear();
jpayne@68: 		}
jpayne@68: 		bf.close();
jpayne@68: 		if(bsw!=null){bsw.poisonAndWait();}
jpayne@68: 	}
jpayne@68: 	
jpayne@68: }