Floating point exception and seg fault in Call Set Refinement with polyclone

[slhogle]

Describe the bug At the stage of "Starting Call Set Refinement (CSR) filtering CSR: Starting registration pass" Octopus crashes with a peculiar error of either "Segmentation Fault" or "Floating point exception." Both error messages have occured using the same command with the same input.

Version

$ octopus --version
octopus version 0.7.4
Target: x86_64 Linux 5.10.25-linuxkit
SIMD extension: AVX2
Compiler: GNU 11.1.0
Boost: 1_76

Command Command line to install octopus:

# Using singularity container pulled from dockerhub
$ singularity build octopus.sif docker://dancooke/octopus

Command line to run octopus:

$ octopus \
--reference HAMBI_2659.fa \
--reads bams/HAMBI_2659-SH-MET-113.bwa.sorted.bam \
--caller polyclone \
--max-clones 5 \
--sequence-error-model PCR.NOVASEQ \
--output HAMBI_2659-SH-MET-113.octopus.vcf \
--annotations AC AD ADP AF AFB DP GQ MP MQ MQ0 \
--forest ${FOREST} \
--threads 1 \
--target-read-buffer-memory=6GB \
--temp-directory-prefix temp 

Additional context I have tried with running many different memory/CPU combinations (on HPC with slurm scheduler) all with the same results. I have run the same command with different BAM files and microbial population genomes with no problems and good results. I am encountering problems when trying to call using reads from metagenomes from a mixed microbial community of 23 known species. My strategy is to do competitive read mapping to all species at once, then partition the big BAM into per-species BAMs, then running Octopus on the single species BAMs. I have also tried to run Ocotpus on the mixed community BAMs always getting the same error.

(I can send BAM and fasta files if required)

octopus_debug.log.gz octopus_trace.log.gz