MISO_SASHIMI step fails when using scratch directory about rnasplice HOT 2 CLOSED

I did a bit more digging -- running it with NXF_DEBUG=1 reveals that two different temporary directories from two different LSF jobs are being used. Specifically, the error is

Exception: Error: no filename /local/xxxxx/20769714/nxf.98A1VB7d5I/index/chrX/ENSG00000005302.pickle

but the job is being executed in /local/xxxxx/20769804/nxf.YYBkdvf6cO:

/bin/bash -c cd /local/xxxxx/20769804/nxf.YYBkdvf6cO; eval export PYTHONNOUSERSITE="1" export R_PROFILE_USER="/.Rprofile" export R_ENVIRON_USER="/.Renviron" export JULIA_DEPOT_PATH="/usr/local/share/julia" export PATH="$PATH:/omics/groups/xxxxx/internal/xxxxx/.nextflow/assets/nf-core/rnasplice/bin"; /bin/bash /omics/odcf/analysis/xxxxx/xxxxx/nextflow/work/33/439d4c44670b38d3bd71734475398f/.command.run nxf_trace

I think the specific problem arises because miso somehow hard-encodes the path in the index. Specifically, if I look for the LSF job ID in the index_gff working directory, I see that

(base) -bash-4.2$ grep -R 20769714 *
Binary file index/genes_to_filenames.shelve.dat matches

so, I suspect that index_gff is hard-encoding the absolute path to the filename. It might be necessary to explicitly specify scratch false for this step. I will try this now.

from rnasplice.

In my hands, setting scratch false in the module seems to make it work (see here). However, is it better to do this in the module, or in conf/modules.config?

from rnasplice.