Minor UG Engine refactoring/cleanup: instead of passing in the # of samples separately from sample set, pass in ploidy instead and compute # of chromosomes internally - will help later on with code clarity

This commit is contained in:
Guillermo del Angel 2012-03-29 11:05:42 -04:00
parent a0843f125e
commit c9c3f6b0fc
2 changed files with 10 additions and 6 deletions

View File

@ -222,7 +222,7 @@ public class UnifiedGenotyper extends LocusWalker<VariantCallContext, UnifiedGen
verboseWriter.println("AFINFO\tLOC\tREF\tALT\tMAF\tF\tAFprior\tMLE\tMAP"); verboseWriter.println("AFINFO\tLOC\tREF\tALT\tMAF\tF\tAFprior\tMLE\tMAP");
annotationEngine = new VariantAnnotatorEngine(Arrays.asList(annotationClassesToUse), annotationsToUse, annotationsToExclude, this, getToolkit()); annotationEngine = new VariantAnnotatorEngine(Arrays.asList(annotationClassesToUse), annotationsToUse, annotationsToExclude, this, getToolkit());
UG_engine = new UnifiedGenotyperEngine(getToolkit(), UAC, logger, verboseWriter, annotationEngine, samples, 2*samples.size()); UG_engine = new UnifiedGenotyperEngine(getToolkit(), UAC, logger, verboseWriter, annotationEngine, samples, UnifiedGenotyperEngine.DEFAULT_PLOIDY);
// initialize the header // initialize the header
Set<VCFHeaderLine> headerInfo = getHeaderInfo(); Set<VCFHeaderLine> headerInfo = getHeaderInfo();

View File

@ -52,6 +52,8 @@ import java.util.*;
public class UnifiedGenotyperEngine { public class UnifiedGenotyperEngine {
public static final String LOW_QUAL_FILTER_NAME = "LowQual"; public static final String LOW_QUAL_FILTER_NAME = "LowQual";
public static final int DEFAULT_PLOIDY = 2;
public enum OUTPUT_MODE { public enum OUTPUT_MODE {
/** produces calls only at variant sites */ /** produces calls only at variant sites */
EMIT_VARIANTS_ONLY, EMIT_VARIANTS_ONLY,
@ -98,7 +100,8 @@ public class UnifiedGenotyperEngine {
private final Logger logger; private final Logger logger;
private final PrintStream verboseWriter; private final PrintStream verboseWriter;
// number of chromosomes (2 * samples) in input // number of chromosomes (ploidy * samples) in input
private final int ploidy;
private final int N; private final int N;
// the standard filter to use for calls below the confidence threshold but above the emit threshold // the standard filter to use for calls below the confidence threshold but above the emit threshold
@ -115,11 +118,11 @@ public class UnifiedGenotyperEngine {
// --------------------------------------------------------------------------------------------------------- // ---------------------------------------------------------------------------------------------------------
@Requires({"toolkit != null", "UAC != null"}) @Requires({"toolkit != null", "UAC != null"})
public UnifiedGenotyperEngine(GenomeAnalysisEngine toolkit, UnifiedArgumentCollection UAC) { public UnifiedGenotyperEngine(GenomeAnalysisEngine toolkit, UnifiedArgumentCollection UAC) {
this(toolkit, UAC, Logger.getLogger(UnifiedGenotyperEngine.class), null, null, SampleUtils.getSAMFileSamples(toolkit.getSAMFileHeader()), 2*(SampleUtils.getSAMFileSamples(toolkit.getSAMFileHeader()).size())); this(toolkit, UAC, Logger.getLogger(UnifiedGenotyperEngine.class), null, null, SampleUtils.getSAMFileSamples(toolkit.getSAMFileHeader()), DEFAULT_PLOIDY*(SampleUtils.getSAMFileSamples(toolkit.getSAMFileHeader()).size()));
} }
@Requires({"toolkit != null", "UAC != null", "logger != null", "samples != null && samples.size() > 0","N>0"}) @Requires({"toolkit != null", "UAC != null", "logger != null", "samples != null && samples.size() > 0","ploidy>0"})
public UnifiedGenotyperEngine(GenomeAnalysisEngine toolkit, UnifiedArgumentCollection UAC, Logger logger, PrintStream verboseWriter, VariantAnnotatorEngine engine, Set<String> samples, int N) { public UnifiedGenotyperEngine(GenomeAnalysisEngine toolkit, UnifiedArgumentCollection UAC, Logger logger, PrintStream verboseWriter, VariantAnnotatorEngine engine, Set<String> samples, int ploidy) {
this.BAQEnabledOnCMDLine = toolkit.getArguments().BAQMode != BAQ.CalculationMode.OFF; this.BAQEnabledOnCMDLine = toolkit.getArguments().BAQMode != BAQ.CalculationMode.OFF;
genomeLocParser = toolkit.getGenomeLocParser(); genomeLocParser = toolkit.getGenomeLocParser();
this.samples = new TreeSet<String>(samples); this.samples = new TreeSet<String>(samples);
@ -130,7 +133,8 @@ public class UnifiedGenotyperEngine {
this.verboseWriter = verboseWriter; this.verboseWriter = verboseWriter;
this.annotationEngine = engine; this.annotationEngine = engine;
this.N = N; this.ploidy = ploidy;
this.N = samples.size() * ploidy;
log10AlleleFrequencyPriorsSNPs = new double[N+1]; log10AlleleFrequencyPriorsSNPs = new double[N+1];
log10AlleleFrequencyPriorsIndels = new double[N+1]; log10AlleleFrequencyPriorsIndels = new double[N+1];
computeAlleleFrequencyPriors(N, log10AlleleFrequencyPriorsSNPs, UAC.heterozygosity); computeAlleleFrequencyPriors(N, log10AlleleFrequencyPriorsSNPs, UAC.heterozygosity);