Why is HaplotypeCaller's '--emitRefConfidence' outputting 'NON_REF' in addition to other ALTs?
Posted in Ask the GATK team | Last updated on 2014-03-18 13:31:09

Comments (11)


When using HaplotypeCaller's '--emitRefConfidence', with both GVCF and BP_RESOLUTION, we get lines like this one: 1 2337032 rs1129171 C T,<NON_REF> 480.77 . BaseQRankSum=0.218;ClippingRankSum=0.103;DB;DP=45;MLEAC=1,0;MLEAF=0.500,0.00;MQ=60.00;MQ0=0;MQRankSum=-1.344;ReadPosRankSum=1.046 GT:AD:DP:GQ:PL:SB 0/1:19,26,0:45:99:509,0,330,565,407,97 2:9,10,13,13

The new 3.0 tool GenotypeGVCFs has no problem processing these lines. However CombineVariants fails to run with this message:

##### ERROR MESSAGE: Reference records should not have more than one alternate allele

By looking at the relevant code:

public/gatk-framework/src/main/java/org/broadinstitute/sting/utils/variant/GATKVariantContextUtils.java if( vc.hasAllele(NON_REF_SYMBOLIC_ALLELE) ) { if( vc.getAlternateAlleles().size() > 1 ) { throw new IllegalStateException("Reference records should not have more than one alternate allele"); }

Sounds like the presence of <NON_REF> in lines with ALTs is triggering this exception.

I also looked at this forum entry and the example there doesn't have <NON_REF> in lines with ALTs.

"Using the reference confidence calculation mode & generating a GVCF"

Should I be able to run CombineVariants on files produced using '--emitRefConfidence'?

Thanks, Carlos PS: I tried to include links to github and GATK forum, but the spam filter seems not to like that.

Return to top Comment on this article in the forum