JexlExpression with VariantEval

pulit

Hello,

I'm currently running variantEval to count up variants per individual stratified by a variety of annotations.

My GATK call looks like:

java -jar /humgen/gsa-hpprojects/GATK/bin/current/GenomeAnalysisTK.jar \ -T VariantEval \ -R Homo_sapiens_assembly19.fasta \ -o output.txt \ -L input.vcf \ -eval input.vcf \ -ST Sample -noST \ -noEV -EV CountVariants \ -ST JexlExpression --select_names "nonsynon" --select_exps "resource.VAT_CDS == 'nonsynonymous' && resource.FOUNDERS_FRQ > 0.05" \ -ST JexlExpression --select_names "synon" --select_exps "resource.VAT_CDS == 'synonymous' && resource.FOUNDERS_FRQ > 0.05" ...

where the VAT_CDS section of the INFO field in the VCF has a functional annotation or is set to "na" if an annotation is unavailable. I'm getting the following error:

ERROR Visit our website and forum for extensive documentation and answers to

ERROR commonly asked questions http://www.broadinstitute.org/gatk

ERROR

ERROR MESSAGE: Invalid command line: Invalid JEXL expression detected for synon with message For input string: "nan"

ERROR ------------------------------------------------------------------------------------------

but weirdly the error is not consistent (my data is stratified by chromosome and most chromosomes will run without throwing the error while one or two chromosomes do exhibit the error. Do you have any ideas what's causing this behavior?

Thanks!

Geraldine_VdAuwera

That is a little weird. Can you confirm that your expression is working on a small subsample of your file? If so, can you narrow down the records that are failing?

pulit

Hi Geraldine,

I can confirm that the expression was working on a subset of the data which would seem to indicate that it was indeed a specific record causing the error.

I tried narrowing down which variant in particular was causing the problem so I reran the script to do this and, lo and behold, did not get an error this time around. Sorry I can't be more helpful here (though I'm glad everything is error free at the moment).

The only difference between the run that caused the error and the one that did not was that I updated the INFO field in the VCF using the variantAnnotator the first time around but the second time wrote my own script to do it. Not sure why that would have changed things.

If I stumble across the error again, I'll try digging into it a bit more.

Sara