gridss.sh error: "Reference genome does not have an index"

[abberclark-diamondage]

I am getting the following error: Reference genome does not have an index. Create using samtools faidx. However, the index file that I generated with samtools faidx is in the same directory as the reference genome.

I am using a custom reference genome in which a transgene has been added to the mouse genome (specifically, GRCm39.primary_assembly.genome.fa).

I get the error using gridss installed via the docker image (gridss/gridss:latest) and using the precompiled version of gridsstools (v2.11.1). I am using an m5.4xlarge EC2 instance (16 vCPU, 64 GiB memory).

This is the command I am running with the precompiled version of gridsstools:
bash gridss.sh --reference /home/ubuntu/Data/GRCm39.primary_assembly.genome_pHM3.fa \ --output /home/ubuntu/Data/SH5588_SA42196.vcf.gz \ --assembly /home/ubuntu/Data/SH5588_SA42196_gridss_assembly.bam \ --threads 8 \ --jar gridss-2.11.1-gridss-jar-with-dependencies.jar \ --workingdir /home/ubuntu/Data/ \ /home/ubuntu/Data/SH5588_SA42196.bam

Here is the full error log.

Please let me know what other information I can provide to help debug the issue.

Thank you!

[d-cameron]

I've been getting a few issues related to this recently and I haven't been able to reproduce locally. Can you post a list of all files in /home/ubuntu/Data/ both before and after the failing GRIDSS run (ls -l /home/ubuntu/Data/)?

[abberclark-diamondage]

Before:

After:

[d-cameron]

Did you manage to resolve this issue? I have only been able to reproduce this locally if the samtools faidx index creation crashes/is killed and the .fai file is empty. Deleting the empty files fixes this issue.