Skip to content

Core Dump Issue during Rare Phasing #117

Open
@ChrisCarson27

Description

@ChrisCarson27

Hi, I am running shapeit5 and coming across a consistent error while using phase_rare. The command looks like something like this repeated for all the chunks of the chromosomes:

phase_rare \
--input /sandbox/PMBB.notrios.rare_prepared.chr22.bcf \
--input-region chr22:37177554-42325255 \
--scaffold /scaffold/PMBB.notrios.phased.chr22.bcf \
--scaffold-region chr22:36677559-43215583 \
--map /maps/chr22.b38.gmap.gz \
--pbwt-mac 2 \
--pbwt-depth-rare 4 \
--pbwt-modulo 0.1 \
--pbwt-mdr 0.1 \
--effective-size 15000 \
--output /output/PMBB.notrios.phased_5.chr22.bcf \
--log /output/PMBB.notrios.phased_5.chr22.log \
--thread 5

Many of the chromosomes phase fine and ligate fine. Some chunks seem to error out with this consistent error. When inspecting the failing regions, nothing stands out. I am using the chunks provided in the github.

The resulting error is typically:

phase_rare: src/containers/genotype_set/genotype_set_phasing.cpp:63: void genotype_set::phaseLiAndStephens(unsigned int, unsigned int, aligned_vector32&, aligned_vector32&, std::vector&, float): Assertion `GRvar_genotypes[vr][tidx].prob >= 0.0f' failed.

The errors result in a core dump.

I have used both the static binaries and an installation that was functional at another lab that was put into a singularity container. The error is consistent across both approaches.

Metadata

Metadata

Assignees

No one assigned

    Labels

    No labels
    No labels

    Projects

    No projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions