@@ -21,7 +21,7 @@ public abstract class SVClusterLinkage<T extends SVLocatable> {
2121 * @param a first item
2222 * @param b second item
2323 */
24- public abstract boolean areClusterable (final T a , final T b );
24+ public abstract LinkageResult areClusterable (final T a , final T b );
2525
2626 /**
2727 * Returns the maximum feasible starting position of any other item with the given item. That is, given item A and
@@ -46,62 +46,56 @@ public int getMaxClusterableStartingPosition(final Collection<T> items) {
4646 }
4747
4848 /**
49- * Checks for minimum fractional sample overlap of the two sets. Defaults to true if both sets are empty.
49+ * Returns number of overlapping items
5050 */
51- protected static boolean hasSampleSetOverlap (final Set <String > samplesA , final Set <String > samplesB , final double minSampleOverlap ) {
52- final int denom = Math .max (samplesA .size (), samplesB .size ());
51+ protected static double getSampleSetOverlap (final Collection <String > a , final Set <String > b ) {
52+ final double denom = Math .max (a .size (), b .size ());
5353 if (denom == 0 ) {
54- return true ;
54+ return 1 ;
5555 }
56- final double sampleOverlap = getSampleSetOverlap (samplesA , samplesB ) / (double ) denom ;
57- return sampleOverlap >= minSampleOverlap ;
56+ return a .stream ().filter (b ::contains ).count () / denom ;
5857 }
5958
6059 /**
61- * Returns number of overlapping items
60+ * Returns true if the overlap is null or exceeds threshold.
6261 */
63- protected static int getSampleSetOverlap (final Collection < String > a , final Set < String > b ) {
64- return ( int ) a . stream (). filter ( b :: contains ). count () ;
62+ protected static boolean testSampleOverlap (final Double sampleOverlap , final double threshold ) {
63+ return sampleOverlap == null || sampleOverlap >= threshold ;
6564 }
6665
66+
6767 /**
68- * Returns true if there is sufficient fractional carrier sample overlap in the two records. For CNVs, returns true
69- * if sufficient fraction of copy number states match.
68+ * Returns fractional carrier sample overlap in the two records. For CNVs, returns fraction of copy number states that match.
7069 */
71- protected static boolean hasSampleOverlap (final SVCallRecord a , final SVCallRecord b , final double minSampleOverlap ) {
72- if (minSampleOverlap > 0 ) {
73- if (a .getType () == GATKSVVCFConstants .StructuralVariantAnnotationType .CNV || b .getType () == GATKSVVCFConstants .StructuralVariantAnnotationType .CNV ) {
74- // CNV sample overlap
75- final GenotypesContext genotypesA = a .getGenotypes ();
76- final GenotypesContext genotypesB = b .getGenotypes ();
77- final Set <String > samples = new HashSet <>(SVUtils .hashMapCapacity (genotypesA .size () + genotypesB .size ()));
78- samples .addAll (genotypesA .getSampleNames ());
79- samples .addAll (genotypesB .getSampleNames ());
80- if (samples .isEmpty ()) {
81- // Empty case considered perfect overlap
82- return true ;
83- }
84- int numMatches = 0 ;
85- for (final String sample : samples ) {
86- final Genotype genotypeA = genotypesA .get (sample );
87- final Genotype genotypeB = genotypesB .get (sample );
88- // If one sample doesn't exist in the other set, assume reference copy state
89- final int cnA = getCopyState (genotypeA , genotypeB );
90- final int cnB = getCopyState (genotypeB , genotypeA );
91- if (cnA == cnB ) {
92- numMatches ++;
93- }
70+ protected static Double computeSampleOverlap (final SVCallRecord a , final SVCallRecord b ) {
71+ if (a .getType () == GATKSVVCFConstants .StructuralVariantAnnotationType .CNV || b .getType () == GATKSVVCFConstants .StructuralVariantAnnotationType .CNV ) {
72+ // CNV sample overlap
73+ final GenotypesContext genotypesA = a .getGenotypes ();
74+ final GenotypesContext genotypesB = b .getGenotypes ();
75+ final Set <String > samples = new HashSet <>(SVUtils .hashMapCapacity (genotypesA .size () + genotypesB .size ()));
76+ samples .addAll (genotypesA .getSampleNames ());
77+ samples .addAll (genotypesB .getSampleNames ());
78+ if (samples .isEmpty ()) {
79+ return null ;
80+ }
81+ int numMatches = 0 ;
82+ for (final String sample : samples ) {
83+ final Genotype genotypeA = genotypesA .get (sample );
84+ final Genotype genotypeB = genotypesB .get (sample );
85+ // If one sample doesn't exist in the other set, assume reference copy state
86+ final int cnA = getCopyState (genotypeA , genotypeB );
87+ final int cnB = getCopyState (genotypeB , genotypeA );
88+ if (cnA == cnB ) {
89+ numMatches ++;
9490 }
95- final int numSamples = samples .size ();
96- return (numMatches / (double ) numSamples ) >= minSampleOverlap ;
97- } else {
98- // Non-CNV
99- final Set <String > samplesA = a .getCarrierSampleSet ();
100- final Set <String > samplesB = b .getCarrierSampleSet ();
101- return hasSampleSetOverlap (samplesA , samplesB , minSampleOverlap );
10291 }
92+ final int numSamples = samples .size ();
93+ return (numMatches / (double ) numSamples );
10394 } else {
104- return true ;
95+ // Non-CNV
96+ final Set <String > samplesA = a .getCarrierSampleSet ();
97+ final Set <String > samplesB = b .getCarrierSampleSet ();
98+ return getSampleSetOverlap (samplesA , samplesB );
10599 }
106100 }
107101
@@ -121,4 +115,13 @@ private static int getCopyState(final Genotype genotype, final Genotype matchedS
121115 VariantContextGetters .getAttributeAsInt (genotype , GATKSVVCFConstants .DEPTH_GENOTYPE_COPY_NUMBER_FORMAT , -1 ));
122116 }
123117 }
118+
119+ /**
120+ * Used for storing the result of a clustering check between two records and any additional metadata
121+ */
122+ public static class LinkageResult {
123+ private final boolean result ;
124+ public LinkageResult (final boolean result ) { this .result = result ; }
125+ public boolean getResult () { return result ; }
126+ }
124127}
0 commit comments