
Laura Gauthier
- Total activity 118
- Last activity
- Member since
- Following 0 users
- Followed by 0 users
- Votes 0
- Subscriptions 49
Activity overview
Latest activity by Laura Gauthier-
Laura Gauthier commented,
Based on the progress in the logs do you expect it to finish? Historically, we've let jobs run for multiple days if they have to, but parallelized into smaller shards if we expect jobs to take mor...
-
Laura Gauthier commented,
Official comment The part of your log that caught my attention is the Model could not pre-compute denominators. I believe this happens when the covariance matrix is not invertible, usually because the variance of o...
-
Laura Gauthier commented,
Ah, the options I'm looking at are for GenotypeGVCFs, so they won't speed up your import. https://gatk.broadinstitute.org/hc/en-us/articles/13832766863259-GenotypeGVCFs. Did you run ReblockGVCF on...
-
Laura Gauthier commented,
Hi Yifei Dai, If you're seeing slowdowns in the middle, it's probably because of the data itself. Can you see the logs while the jobs are running? If what I think is happening is true, then there...
-
Laura Gauthier commented,
Hi Rebeka Bató At this point, we don't have a dedicated GATK support team anymore and so we don't have the capacity to critique anyone's experimental design. That said, since you did enumerate you...
-
Laura Gauthier commented,
SplitIntervals will divide the calling intervals, but all the shards will get the same data. They'll just read and combine what they need. The splitting should be quick and easy.
-
Laura Gauthier commented,
Hi Yifei Dai, To be perfectly explicit, are you using the batch size argument when running GenomicsDB? We like to use 50. Having 600 files open at the same time (if that's the actual size of your...
-
Laura Gauthier commented,
Hi Emila Mańko I've seen a decrease in the max gaussians argument help when the error is no negative training data, but that's not what your log shows. It's going to be hard to further debug witho...
-
Laura Gauthier commented,
Hi Emila Mańko, It's suspicious to me that it says zero variants processed and I have a theory. Can you confirm that there is at least one variant in your training VCF that is also in your input V...
-
Laura Gauthier commented,
Hi xin cui, GATK doesn't do any sort of Bayesian phasing so it's actually quite strict. In order to phase two variants they need to occur on either 100% or 0% of the reads that overlap the two pos...