Open creisle opened 2 years ago
Results
Not unexpected since the pyfaidx will use the index where it exists and biopython presumably is not. Will also need to test how much impact this has on an overall runtime
Seems like there's also some performance advantages to pyfaidx
compared to the samtools version
So there definitely ARE performance boosts but I did run into an annoying issue. As far as I can tell you can't read a fasta without an index. Mostly this would be fine except that if you don't have write permissions to the place where the reference genome lives you can't build one either. Not sure how to resolve this so I am putting this on hold for now
Loading FASTA files with biopython is quite sloe (~30s). Test whether we can replace the biopython load with using the pyfaidx library instead
I copied the current function that uses biopython and modified it to use pyfaidx and then tested loading various reference genome fasta files (3x replication) over 6 human genomes (hg38, hg19. h18, no alt, etc)