support for non-AS called data (#6975)

* support for non-AS called data * fixed null case
broadinstitute · Nov 21, 2020 · 5f40379 · 5f40379
1 parent 2610b20
commit 5f40379
Show file tree

Hide file tree

Showing 2 changed files with 56 additions and 5 deletions.
diff --git a/...ain/java/org/broadinstitute/hellbender/tools/variantdb/nextgen/ExtractFeaturesEngine.java b/...ain/java/org/broadinstitute/hellbender/tools/variantdb/nextgen/ExtractFeaturesEngine.java
@@ -19,7 +19,6 @@
 import org.broadinstitute.hellbender.tools.walkers.annotator.VariantAnnotatorEngine;
 import org.broadinstitute.hellbender.tools.walkers.annotator.allelespecific.AS_StrandBiasTest;
 import org.broadinstitute.hellbender.utils.QualityUtils;
-import org.broadinstitute.hellbender.utils.SimpleInterval;
 import org.broadinstitute.hellbender.utils.bigquery.BigQueryUtils;
 import org.broadinstitute.hellbender.utils.bigquery.GATKAvroReader;
 import org.broadinstitute.hellbender.utils.bigquery.StorageAPIAvroReader;
@@ -130,6 +129,11 @@ private void processVQSRRecordForPosition(GenericRecord rec) {
         String ref = rec.get("ref").toString();
         String allele = rec.get("allele").toString();
 
+        if (allele == null || allele.equals("")) {
+            logger.warn("SEVERE WARNING: skipping " + contig + ":" + position + "(location="+location+") because it has a null alternate allele!");
+            return;
+        }
+
         // Numbers are returned as Long (sci notation)
         Double qual = Double.valueOf(rec.get(SchemaUtils.RAW_QUAL).toString());
 
@@ -147,10 +151,10 @@ private void processVQSRRecordForPosition(GenericRecord rec) {
         Double raw_ad_gt_1 = Double.valueOf(rec.get("RAW_AD_GT_1").toString());
 
         // TODO: KCIBUL QUESTION -- if we skip this... we won't have YNG Info @ extraction time?
-        if (raw_ad == 0) {
-            logger.info("skipping " + position + " because it has no alternate reads!");
-            return;
-        }
+        // if (raw_ad == 0) {
+        //     logger.info("skipping " + contig + ":" + position + "(location="+location+") because it has no alternate reads!");
+        //     return;
+        // }
 
         int sb_ref_plus = Double.valueOf(rec.get("SB_REF_PLUS").toString()).intValue();
         int sb_ref_minus = Double.valueOf(rec.get("SB_REF_MINUS").toString()).intValue();

diff --git a/src/main/java/org/broadinstitute/hellbender/tools/variantdb/nextgen/VetFieldEnum.java b/src/main/java/org/broadinstitute/hellbender/tools/variantdb/nextgen/VetFieldEnum.java
@@ -121,7 +121,31 @@ public String getColumnValue(final VariantContext variant) {
 
     AS_RAW_MQRankSum { // TODO -- maybe rely on 1/1 for call_GT, also get rid of the | at the beginning
         public String getColumnValue(final VariantContext variant) {
+            // e.g. AS_RAW_MQRankSum=|1.4,1|NaN;
             String out =  getAttribute(variant, GATKVCFConstants.AS_RAW_MAP_QUAL_RANK_SUM_KEY, null);
+
+            if (out == null) {
+                // Try to use non-AS version
+                // TODO: it looks like the AS_RAW version also trims to a single decimal point??
+                // e.g. MQRankSum=1.465 and turn it into |1.465,1|
+                String outNotAlleleSpecific = getAttribute(variant, GATKVCFConstants.READ_POS_RANK_SUM_KEY, null);
+
+
+                if ( outNotAlleleSpecific == null || "".equals(outNotAlleleSpecific) || outNotAlleleSpecific.contentEquals("||") || outNotAlleleSpecific.contentEquals("|||") ) {
+                    return "";
+                }
+
+                if (variant.getAlleles().size() == 3) { // GT 0/1 1/1
+                    out = "|" + outNotAlleleSpecific + ",1|";
+
+                } else if (variant.getAlleles().size() == 4) { // GT 1/2
+                    // TODO: just replicate rather than distribute, is this right?
+                    out = "|" + outNotAlleleSpecific + ",1|" + outNotAlleleSpecific + ",1|";
+                } else {
+                    throw new UserException("Expected diploid sample to either have 3 alleles (ref, alt, non-ref) or 4 alleles (ref, alt 1, alt 2, non-ref)");
+                }               
+            }
+
             if ( out == null || out.contentEquals("||") || out.contentEquals("|||") ) {
                 out = "";
                 return out;
@@ -176,7 +200,30 @@ public String getColumnValue(final VariantContext variant) {
 
     AS_RAW_ReadPosRankSum {  // TODO -- maybe rely on 1/1 for call_GT
         public String getColumnValue(final VariantContext variant) {
+            // e.g. AS_RAW_ReadPosRankSum=|-0.3,1|0.6,1
             String out =  getAttribute(variant, GATKVCFConstants.AS_RAW_READ_POS_RANK_SUM_KEY, null);
+
+            if (out == null) {
+                // Try to use non-AS version
+                // TODO: it looks like the AS_RAW version also trims to a single decimal point??
+                // e.g. ReadPosRankSum=-0.511 and turn it into |-0.511,1|
+                String outNotAlleleSpecific = getAttribute(variant, GATKVCFConstants.READ_POS_RANK_SUM_KEY, null);
+
+                if ( outNotAlleleSpecific == null || "".equals(outNotAlleleSpecific) || outNotAlleleSpecific.contentEquals("||") || outNotAlleleSpecific.contentEquals("|||") ) {
+                    return "";
+                }
+
+                if (variant.getAlleles().size() == 3) { // GT 0/1 1/1
+                    out = "|" + outNotAlleleSpecific + ",1|";
+
+                } else if (variant.getAlleles().size() == 4) { // GT 1/2
+                    // TODO: just replicate rather than distribute, is this right?
+                    out = "|" + outNotAlleleSpecific + ",1|" + outNotAlleleSpecific + ",1|";
+                } else {
+                    throw new UserException("Expected diploid sample to either have 3 alleles (ref, alt, non-ref) or 4 alleles (ref, alt 1, alt 2, non-ref)");
+                }               
+            }
+
             if (out == null || out.contentEquals("||") || out.contentEquals("|||") ) {
                 out = "";
                 return out;