Segmentation fault caused by bustools_correct_sort

[nf-core]

Segmentation fault caused by bustools_correct_sort

[apeltzer]

I double checked this behaviour and cannot further adjust this atm - we already use the -m parameter and set the value to whatever the profile specifies when running the process (depending on the host system / profile configuration for that task.) See:

scrnaseq/main.nf

Line 639 in 884e541

bustools sort -T tmp/ -t ${task.cpus} -m ${task.memory.toGiga()}G -o ${bus}/output.corrected.sort.bus $sort_file

Not sure we can further improve that, I'm wondering whether this is the real reason 🤔

[galanisl]

I noticed that the intermediate output.bus files generated by kallisto were empty (see Read in 0 BUS records above) and this causes the segmentation fault in the bustools sort step. My fastq files had the following permissions:

u=rwx,g=,o=

Changing them to u=rwx,g=rx,o=r solved the problem in my HPC environment.