Thanks for sharing your analysis. I plan on using this in the new year, and noticed that running the pre-processing scripts on our shared HPC cluster resulting in some permissions issues for shared reference genome assemblies. The issue is that pyfasta creates its own index sidecar files, and fails if the user cannot write to the shared resource directory. I've swapped out pyfasta for my pyfaidx module, which will read or create a samtools .fai index file, which is likely already present in such a scenario. This also allowed me to drop the biopython dependency, as pyfaidx has a FASTA wrapping function.
The output on the example dataset is identical using the code in this PR vs the current master branch.
Thanks for sharing your analysis. I plan on using this in the new year, and noticed that running the pre-processing scripts on our shared HPC cluster resulting in some permissions issues for shared reference genome assemblies. The issue is that
pyfasta
creates its own index sidecar files, and fails if the user cannot write to the shared resource directory. I've swapped out pyfasta for my pyfaidx module, which will read or create a samtools .fai index file, which is likely already present in such a scenario. This also allowed me to drop the biopython dependency, as pyfaidx has a FASTA wrapping function.The output on the example dataset is identical using the code in this PR vs the current master branch.