The current GATK version is 3.3-0

Howdy, Stranger!

It looks like you're new here. If you want to get involved, click one of these buttons!

HaplotypeCaller LikelihoodCalculationEngine.computeDiploidHaplotypeLikelihoods NullPointerException

LondonPosts: 19Member

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

Any help would be great.

Thanks

Andrew

• UKPosts: 4Member

hello Geraldine, What file(s) should I send/attach/link?

0) original log file(s)

1) [chr14 part (where HC crashed) of] Bam file

2) [chr14 part of] ref fasta file (.fa.gz)

3) [chr14 part of] known variant file (.vcf.gz)

4) [chr14 part of] variant output file(s) up until (.vcf.gz)

thanks, Mikyung

• UKPosts: 4Member

hello Geraldine and team, While I was waiting for the answers, I tried my third run of HC and it finished this time. This time the estimated runtime stayed >47h and finished after 50.1h. At the two previous failed instances, estimated runtime were 29h and 33h each and died at 16h and 18.4h each. I might run another round of indel realignment to variant calling. I'll be back if I have error again.

thanks, Mikyung

• SeattlePosts: 3Member

@mjang said: hello Geraldine and team, While I was waiting for the answers, I tried my third run of HC and it finished this time. This time the estimated runtime stayed >47h and finished after 50.1h. At the two previous failed instances, estimated runtime were 29h and 33h each and died at 16h and 18.4h each. I might run another round of indel realignment to variant calling. I'll be back if I have error again.

thanks, Mikyung

thanks for posting this... i'm going to re-run my hc analysis too..

did you change anything before your third run?

It seems that these are further instances of a transient bug that we haven't yet nailed down because we can't reproduce it locally. To deal with this, we recommend scatter-gathering your jobs so that if one region fails due to the bug, you can just restart that and not the entire job. You can do this failry easily with Queue.

Geraldine Van der Auwera, PhD

• SeattlePosts: 3Member

@Geraldine_VdAuwera said: It seems that these are further instances of a transient bug that we haven't yet nailed down because we can't reproduce it locally. To deal with this, we recommend scatter-gathering your jobs so that if one region fails due to the bug, you can just restart that and not the entire job. You can do this failry easily with Queue.

• LondonPosts: 19Member

I ran the HC without threading (no -nct or nt) and splitting up the file based on my exome sequenced regions and running those in parallel (without using Queue at the moment) and with that setup I didn't get any exceptions. Might be worth putting a note against the HC -nt option that it is buggy?

Thanks

Tim