ugly, but it works: I can run a maximum of 50 jobs with a maximum of 500 cpus on our server.

So I run 10 # gatk HaplotypeCaller in parallel in one nextflow process using `make -j 10` ===> 500 gatk HC at the same time.

https://gitlab.univ-nantes.fr/pierre.lindenbaum/gazoduc-nf/-/blob/master/modules/gatk/hapcaller2/main.nf?ref_type=heads#L81

#nextflow #gatk #make #makefile #bioinformatics #workflow

https://github.com/WeTheAgents/markdup8x-wea
"
markdup-wea: Rust rewrite of Picard MarkDuplicates — 3.26x faster, 54x less RAM, byte-identical output on 8 ENCODE RNA-seq samples vs Picard 3.4.0. Plus nf-core/rnaseq Picard memory-tuning proposal"

#bioinformatics #gatk #rust

I asked #biostars : "Parabricks : Number of GPUs requested (2) is more than number of GPUs (0) in the system., exiting."
https://www.biostars.org/p/9614740/

#bioinformatics #parabricks #NVIDIA #haplotypecaller #gatk #gpus

Know what would have been cool? If Broad hadn't broken two thirds of the links in the GATK documentation when they redid the website.

#BroadInstitute #GATK

https://xcancel.com/gatk_dev/status/1908902547521257779 "GATK Dev Team :" GATK forums will be community-driven and self-moderated. They will not be moderated or monitored by a GATK team member. We encourage members of the community to continue to engage with each other on these forums." https://gatk.broadinstitute.org/hc/en-us/community/topics

#gatk

TIL: You can get a list of Spark-enabled GATK tools with the command

gatk --list | grep Spark

(The website doesn't seem to have a list anywhere)

#bioinformatics #GATK #ApacheSpark

So after much community backlash #GATK has reverted back to the ./. representation for no-calls.

The question is, will something like this occur again? and how many people have already switched away to other tools like #bcftools.

I know I have moved all my pipelines back to bcftools and honestly its brilliant.

#genomics #bioinformatics

https://github.com/broadinstitute/gatk/releases/tag/4.6.0.0