zzh
/
gatk-3.8
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

Thanks to mark: VCFInfoToTable removed in favor of a more flexible walker. Slight change to the argument structure of the walker to make it play more nicely with Queue: the field list parsing is pushed into the command line system (e.g. the variable is exposed as a List<String> and not a String, so Queue doesn't have to join a list into a string only to have it broken out again. This also allows the user to specify -F field1 -F field2 -F field3 if he/she so desires. 

git-svn-id: file:///humgen/gsa-scr1/gsa-engineering/svn_contents/trunk@4842 348d0f76-0448-11de-a6fe-93d51630548a
This commit is contained in:
chartl 2010-12-15 03:33:36 +00:00
parent 2217837845
commit 3e75431bc8
3 changed files with 15 additions and 69 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

6
java/src/org/broadinstitute/sting/gatk/walkers/variantutils/VariantsToTable.java
View File

@ -47,7 +47,7 @@ public class VariantsToTable extends RodWalker<Integer, Integer> {
protected PrintStream out;
@Argument(fullName="fields", shortName="F", doc="Fields to emit from the VCF, allows any VCF field, any info field, and some meta fields like nHets", required=true)
public String FIELDS;
public List<String> fieldsToTake = null;
@Argument(fullName="showFiltered", shortName="raw", doc="Include filtered records")
public boolean showFiltered = false;
@ -59,11 +59,7 @@ public class VariantsToTable extends RodWalker<Integer, Integer> {
@Argument(fullName="ignoreMultiAllelic", shortName="IMA", doc="If provided, we will not require the site to be biallelic", required=false)
public boolean ignoreMultiAllelic = false;
private List<String> fieldsToTake;
public void initialize() {
fieldsToTake = Arrays.asList(FIELDS.split(","));
out.println(Utils.join("\t", fieldsToTake));
}

22
scala/qscript/lib/Vcf2Table.q
View File

@ -1,7 +1,6 @@
import org.broadinstitute.sting.commandline.Hidden
import org.broadinstitute.sting.queue.extensions.gatk.{RodBind, VariantsToTable}
import org.broadinstitute.sting.queue.QScript
import org.broadinstitute.sting.queue.library.ipf.vcf.VCFInfoToTable
import collection.JavaConversions._
class Vcf2Table extends QScript {
@ -9,14 +8,21 @@ class Vcf2Table extends QScript {
@Argument(shortName="f",doc="Info fields to extract",required=false) var fields : java.util.List[String] = new java.util.ArrayList[String]
@Argument(shortName="o",doc="Output file",required=true) var output : File = _
@Argument(shortName="useFilters",doc="Use filtered sites?",required=false) var useFilters : Boolean = false
@Hidden @Argument(shortName="pass",doc="set the hack filter string to this value",required=false) var filterString : String = "PASS"
@Hidden @Argument(shortName="notfound",doc="set the hack no entry string to this value",required=false) var keyNotFound : String = "NA"
@Argument(shortName="r",doc="Reference file") var ref : File = _
@Argument(shortName="i",doc="Intervals",required=false) var ints : java.util.List[File] = new java.util.ArrayList[File]
@Argument(shortName="g",doc="gatk jar",required=true) var gatk: File = _
def script = {
var vcf2table : VCFInfoToTable = new VCFInfoToTable(vcf,output,fields,useFilters)
vcf2table.PF_KEY = filterString
vcf2table.NO_KEY = keyNotFound
var vcf2table : VariantsToTable = new VariantsToTable
vcf2table.rodBind :+= new RodBind("variant","vcf",vcf)
vcf2table.reference_sequence = ref
vcf2table.intervals = ints.toList
vcf2table.raw = useFilters
vcf2table.out = output
vcf2table.F = fields.toList
vcf2table.jarFile = gatk
add(vcf2table)
this.functions.foreach(u => logger.debug("added: %s%n".format(u.toString)))
}
}

56
scala/src/org/broadinstitute/sting/queue/library/ipf/vcf/VCFInfoToTable.scala
View File

@ -1,56 +0,0 @@
package org.broadinstitute.sting.queue.library.ipf.vcf
import org.broadinstitute.sting.commandline._
import org.broadinstitute.sting.utils.text.XReadLines
import collection.JavaConversions._
import java.io.{PrintStream, PrintWriter, File}
import collection.immutable.HashSet
import collection.mutable.HashMap
import org.broadinstitute.sting.queue.function.InProcessFunction
class VCFInfoToTable(vcf: File, table: File, annots: List[String], keepFilter: Boolean) extends InProcessFunction {
def this(in: File, out: File, anns: List[String]) = this(in,out,anns,true)
def this(in: File, out: File) = this(in,out,Nil,true)
def this(in: File) = this(in, new File(in.getAbsolutePath.replace(".vcf",".info.table")), Nil, true)
def this(in: File, anns: List[String]) = this(in, new File(in.getAbsolutePath.replace(".vcf",".info.table")),anns,true)
def this(in: File, anns: java.util.List[String]) = this(in, new File(in.getAbsolutePath.replace(".vcf",".info.table")),anns.toList,true)
def this(in: File, anns: List[String], keep: Boolean) = this(in, new File(in.getAbsolutePath.replace(".vcf",".info.table")),anns,keep)
def this(in: File, anns: java.util.List[String], keep: Boolean) = this(in, new File(in.getAbsolutePath.replace(".vcf",".info.table")),anns.toList,keep)
def this(in: File, out: File, anns: java.util.List[String], keep: Boolean) = this(in,out,anns.toList,keep)
@Input(doc="VCF file from which to extract annotion") var inVCF: File = vcf
@Output(doc="Table file to which to write") var outTable: File = table
@Argument(doc="Annotations to extract from info field") var annotations: List[String] = annots
@Argument(doc="Keep filtered records?") var keepFilteredRecs: Boolean = keepFilter
// set as vars so pipelnes can hack these values
var NO_KEY : String = "NA"
var PF_KEY : String = "PASS"
var out : PrintWriter = _
var annotation_set : HashSet[String] = new HashSet[String]
//todo -- Khalid: Why is run not being called?
def run = {
logger.debug("RUN IS CALLED")
annotation_set ++= annotations
out = new PrintWriter(new PrintStream(outTable))
asScalaIterator(new XReadLines(inVCF)).foreach(lineToTable)
}
def lineToTable(line : String) = {
if ( ! line.startsWith("#") ) {
val spline = line.split("\t")
if ( spline(6).equals(PF_KEY) || keepFilteredRecs ) {
val iMap = spline(7).split(";").map(_.split("=")).filter(p => annotation_set.contains(p.apply(0))).foldLeft(new HashMap[String,String])( (a,b) => a += new Tuple2(b.apply(0),b.apply(1)) )
out.print("%s%n".format(annotations.map( u => {
if ( iMap.contains(u) ) {
iMap.get(u)
} else {
NO_KEY
}
}).reduceLeft((a,b) => a + "\t" + b)))
}
}
}
}