cannot find an open port. For manually specifying the port, see ?SnowParamUsing previously downloaded VCF.
#113
Labels
bug
Something isn't working
1. Bug description
import_sumstats
: Works fine for 100s of GWAS, then encounters this error and quickly iterates through all remaining GWAS ids without actually processing them (and, strangely, appends their log files to that of the one that first encountered the error!).This takes a very long time to actually reproduce (multiple days of running continuously). And it's not like the GWAS that was being analyzed at the time of the error was particular large or anything ("only" 11M SNPs).
Possible explanations
BiocParallel
can't handle this gracefully?data.table
is trying to run in parallel within each loop ofread_vcf_parallel
(which is also being run in parallel), causing a conflict with the same cores being requested for different tasks at once. Though I don't know why this wouldn't happen far earlier when processing 100s of GWAS.read_vcf_parallel
:It seems to occur at
read_vcf_parallel
. This function seems to be rather finicky as it also doesn't like it when I specify >30 threads, though I suspect that's for a different reason (splitting a VCF across too many threads means that if some genome tiles are empty, the whole loop breaks, perhaps at the final re-merging step).Related Issues
BiocParallel
:Also, not sure if I'm the only one, but
BiocParallel
can be a bit trickier to use successfully.Console output
Full logs file:
ubm-a-129_log_msg.txt
Expected behaviour
Process all sumstats.
2. Reproducible example
Code
3. Session info
The text was updated successfully, but these errors were encountered: