Closed vodkatad closed 2 months ago
Hi, have you solver the error? I met the same issue with the conda installed optitype
We ended up finding that indeed it was a RAM issue - the aligner razers3 had not enough ram (250Gb) to run with both R1 and R2 reads from 30x WGS. We ended up using R1 and R2 separately, apparently (but I am not sure) optitype does not use the information from read pairs so it should be fine (even if a single allele ended up having different calls with R1/R2, but they are very similar alleles...).
Thank you for your reply. When I run it on the example data, razers3 used more 100G RAM. and then I got that error. Do you think bwa-mem can be used instead of razers3 for the alignament?
We ended up using bwa with these parameters:
bwa mem -M -t $(nproc) $HLA_REF $R1 2> err.bwa-hla.$OUT | \
samtools view -F4 -C -T $HLA_REF -@10
OK, thank you so much!
HLA-Locus prefiltering helps. And yes Optitype uses the paired information when constructing the binary hit matrix
After managing to run the test via docker we've tried with our data, but without success:
RAM should not be an issue. Can anyone give us pointers? Thanks, E.