- Last Active
- Full Name
- Noah Snyder-Mackler
I doubled the memory to 32G and it seemed to work on samples where it failed previously. Thanks for the quick responses. I really appreciate the quick and thoughtful support offered by the GATK team!
HI Sheila, Thanks for letting me know. I'll see if that is the issue. Stay tuned... Noah
(Quote) I thought that might be the case, but it doesn't seem to hold since the recal table generation and printreads are two lines in the same shell script (and therefore part of the same job). I'll see if I can increase the memory footprint and f…
(Quote) Thanks. I guess this makes some sense. I'll need to look into it a bit more. It looks like I should filter by genotypes rather than just by variant site. More post-GATK processing work for me! (Quote) It's a variant site filter, not genotyp…
So it wouldn't work with anything higher than -nct 1, which is too bad because It took 12-36hrs per sample, rather than <12hrs each. However, I ran them in parallel so it wasn't a big deal. Thanks for the helpful feedback!
(Quote) Nope. I'm running java 1.7 They're running just fine with -nct 1 right now, but obviously they are taking much longer...
I've tried it with -nct 4, -nct 2 and now with -nct 1. I also upped the java heap to 16GB. I'm running 24 samples at once, but they are small bam files (low coverage) and I'm running them on 8 high memory nodes on our cluster, so this shouldn't real…
Ok. I will try using just one thread, which will slow everything down, right? I was able to use -nct 32 while running BQSR with java heap Xmx4g and did not run into any problems.