Bug in creation outputvcf file after Beagle

Tiphaine

Hi,

I used Beagle to phase my data but for some indels, I have some probleme :

example :

Input vcf :

2       68599872        .       ATG     A       14.40   PASS    AC=1;AC1=1;AF=0.028

Input for beagle created by ProduceBeagleInput:

2:68599872 TG - 1.0000 0.0000 0.0000 ......

Output vcf created by BeagleOutputToVCF:

2       68599872        .       ATG     .       14.40   BGL_RM_WAS_-    AC1=1;AF1=0.02965.....

error message by CombineVariants:

MESSAGE: Badly formed variant context at location 68599872 in contig 2. Reference length must be at most one base shorter than location size

Can you help me?

Tipahine

Geraldine_VdAuwera

Have you validated the vcf file output by Beagle? If it fails you may need to contact the authors of Beagle -- if their tool is producing bad vcf files, we can't help with that. But if the problem is on our end we'll do what we can.

ebanks

Actually, this looks like it may be a bug in our code. We'll take a quick look and get back to you with some feedback.

Geraldine_VdAuwera

Yeah, a shot of caffeine later I realized our BeagleOutputToVCF may be the culprit, since that's what's making the VCF. Sorry about that.

ebanks

Okay, it looks like Beagle claimed that your site was monomorphic so BeagleOutputToVCF is filtering your site and setting the ALT allele to "." (not polymorphic). This looks reasonable. So the problem you are getting must be in CombineVariants. Are you using the latest version of the GATK? If so, what is your command-line? (And if not, please update to the latest version)

Tiphaine

Sorry, it is not the lastest version but v2.0-35-g2d70733 my command line is : java -jar $GATK_HOME/GenomeAnalysisTK.jar -R $RefGen -T BeagleOutputToVCF -V $VcfFile -beagleR2:BEAGLE $r2 -beaglePhased:BEAGLE $phase -beagleProbs:BEAGLE $probs -o $vcfBeagle -U LENIENT_VCF_PROCESSING -et NO_ET -K $GATK_KEY

ebanks

Sorry, it's the Combine Variants command-line that we need.

Tiphaine

I use it in PBS script so before that, I define value for each variable

 java -jar $GATK_HOME/GenomeAnalysisTK.jar -R $RefGen -T CombineVariants -U LENIENT_VCF_PROCESSIN
 G --out $outputFile -V:input1 $input1 -V:input2 $input2 -V:input3 $input3 -V:input4 $input4 -V:input5 $input5 -V:input
 6 $input6 -V:input7 $input7 -V:input8 $input8 -V:input9 $input9 -V:input10 $input10 -V:input11 $input11 -V:input12 $in
 put12 -V:input13 $input13 -V:input14 $input14 -V:input15 $input15 -V:input16 $input16 -V:input17 $input17 -V:input18 $
 input18 -V:input19 $input19 -V:input20 $input20 -V:input21 $input21 -V:input22 $input22 -V:inputX $inputX -genotypeMer
 geOptions PRIORITIZE -priority input1,input2,input3,input4,input5,input6,input7,input8,input9,input10,input11,input12,
 input13,input14,input15,input16,input17,input18,input19,input20,input21,input22,inputX -et NO_ET -K $GATK_KEY

ebanks

I would try it with the latest version of the GATK. If it still fails, then I recommend trying to find a subset of the data with which you can replicate this error (i.e. just 2 of your input VCF files to CombineVariants) and then post the records at 68599872 here so we can help you figure out where the problem is.