Closed cinquin closed 5 years ago
Thanks for the bug report. The problem is that the operating system maximum open file limit is being exceeded (i.e. ulimit -n
). When calling with multiple threads, Octopus creates a temporary BCF file for each unique contig in the calling region set. Since you're not explicitly specifying any input regions in your command, the default behaviour is it to call all contigs in the reference genome. That's all 3366 contigs in GRCh38_full_analysis_set_plus_decoy_hla.fa
, even though reads are only mapped to chr1
.
A robust solution to this is not going to be simple. However, it's fairly easy for you to avoid this problem by either:
ulimit -n 10000
).-T chr1
to your command).For my own information, some ideas to help - but not solve - this problem:
Thanks; adding -T chr1
does work around the problem.
From 42fa364aa3ca64ea25613458c8f6a45dbab5f34f, Octopus will open and close temporary BCFs as needed, which should reduce the occurrence of this type of error. There is also better error reporting. You should now be able to run your command without the -T chr1
(although I'd still recommend including it if you're only calling chromosome 1).
Describe the bug The program fails early when run in multithreaded mode (both for debug and release builds). Note that the input BAM file only has reads mapped to chromosome 1. Stack trace of the crashed thread below for the debug build.
When changing the
--thread
argument to--thread 1
, the problem does not occur.Command Command line to run octopus:
Desktop (please complete the following information):
OS: macOS 10.14.3
Version commit 785bfa2d06a997a15ad0b38abc31745283e5b4f7 Author: Daniel Cooke dcooke@well.ox.ac.uk Date: Sat Jan 26 22:35:05 2019 +0000
Reference ftp://ftp.1000genomes.ebi.ac.uk/vol1/ftp/technical/reference/GRCh38_reference_genome/GRCh38_full_analysis_set_plus_decoy_hla.fa