Running into errors with conda version while docker/singularity work fine

[famosab]

I am working on a module for nf-core (nf-core/modules#5932) where we are now porting to a new testing framework. Running these tests means giving the same input to the conda, docker and singularity version of the same tool. While both docker and singularity complete without errors, I always run into the following error when running gridss somaticfilter:

Attaching package: ‘Matrix’

The following object is masked from ‘package:S4Vectors’:

    expand

The following objects are masked from ‘package:tidyr’:

    expand, pack, unpack


Attaching package: ‘MatrixGenerics’

The following objects are masked from ‘package:matrixStats’:

    colAlls, colAnyNAs, colAnys, colAvgsPerRowSet, colCollapse,
    colCounts, colCummaxs, colCummins, colCumprods, colCumsums,
    colDiffs, colIQRDiffs, colIQRs, colLogSumExps, colMadDiffs,
    colMads, colMaxs, colMeans2, colMedians, colMins, colOrderStats,
    colProds, colQuantiles, colRanges, colRanks, colSdDiffs, colSds,
    colSums2, colTabulates, colVarDiffs, colVars, colWeightedMads,
    colWeightedMeans, colWeightedMedians, colWeightedSds,
    colWeightedVars, rowAlls, rowAnyNAs, rowAnys, rowAvgsPerColSet,
    rowCollapse, rowCounts, rowCummaxs, rowCummins, rowCumprods,
    rowCumsums, rowDiffs, rowIQRDiffs, rowIQRs, rowLogSumExps,
    rowMadDiffs, rowMads, rowMaxs, rowMeans2, rowMedians, rowMins,
    rowOrderStats, rowProds, rowQuantiles, rowRanges, rowRanks,
    rowSdDiffs, rowSds, rowSums2, rowTabulates, rowVarDiffs, rowVars,
    rowWeightedMads, rowWeightedMeans, rowWeightedMedians,
    rowWeightedSds, rowWeightedVars


Attaching package: ‘S4Arrays’

The following object is masked from ‘package:abind’:

    abind

The following object is masked from ‘package:base’:

    rowsum

No reference genome supplied using --ref. Not performing variant equivalence checks.
2024-09-03 17:14:56.3009 Reading test.vcf.gz
Tumour samples:
Matched normals:
Error: array-like subscript has more than one effective dimension
Execution halted

I am thankful for any hints what could cause this problem.

Command used:

#!/bin/bash -ue
gridss_somatic_filter \
    --input test.vcf.gz \
     \
    --output test.high_confidence_somatic.vcf \
    --fulloutput test.all_somatic.vcf \
    --normalordinal 0

cat <<-END_VERSIONS > versions.yml
"GRIDSS_SOMATICFILTER":
    gridss: 2.13.2
END_VERSIONS

[lwtan90]

I am trying to understand normalordinal parameter. The default was 1. Is there a reason why your command involves --normalordinal 0 ?

[famosab]

I am trying to understand normalordinal parameter. The default was 1. Is there a reason why your command involves --normalordinal 0 ?

I'm afraid I cannot help you there. I did not configure it that way.
I think 0 just means that we do not have a matched normal in that case.

[SPPearce]

The exact same issue happens when--normalordinal 0.
This test is using the output of GRIDSS on a tiny test dataset that doesn't include any SVs, so the vcf file is empty of content (just has the headers).
Presumably there is a difference in an underlying R package when run through conda in this case, which leads to the error above, whereas Docker/Singularity don't complain and produce valid empty output.

Do you have a valid gridss vcf file that could be used as test-data?

[SPPearce]

I have just tried on an actual GRIDSS output vcf file (or the first 500 lines of it anyway).
This worked fine with the bioconda docker/singularity container, but a new one (built via seqera containers) is failing as well.
So there is a more general issue with the current bioconda recipe.