1
1
Entering edit mode
@jonathan-griffiths-20559
Last seen 4 months ago
Cambridge UK

Hello,

I was mapping some poorly annotated RNA-seq data using subread and incorrectly understood it to be paired end data.

Supplying the fastq files as PE reads to subread-align caused a near-instant segfault. Should it provide a clearer error message?

I can supply some offending files if you like, but I suspect this behaviour will be reproducible with any data you have lying around.

Thanks.

E: just to add, this is on version 1.6.4, the command would have been something very much like: subread-align -r $file -R$file2 -i $index -t 0 -P 3 -T 1 -o$out. It wasn't very clear to the user exactly why this segfault was happening, either, hence my suggestion here.

3
Entering edit mode
@gordon-smyth
Last seen 9 hours ago
WEHI, Melbourne, Australia

I do not get a segmentation error from the Bioconductor package Rsubread (current devel version). Running

align(hg38, file1, file2)


where hg38 is the index path and file1 and file2 are single-end RNA-seq FastQ files for different samples, the programs simply runs through the files until it reaches the end of the shorter of the two files, then it outputs an informative error message:

ERROR: two input files have different amounts of reads!
The program has to terminate and no alignment results were generated!


I haven't tried the command-line version of Subread but the C code is the same as for the R package. I can't see how you could get a "near instant" error unless something was seriously wrong with your subread index, because the program has to load the index before it start processing the FastQ files, and that in itself takes a minute or two.

0
Entering edit mode

Indeed, regenerating my index fixed this (though there was no apparent error while it was being made). Thanks for the clarity here!