Open jrkirk61 opened 1 month ago
Update: It turns out that all 200k reads did not align to the genome, which was a very unexpected result. It seems that minimap2 saves everything including unaligned reads so it's hard to tell this based on the size of the BAM file alone. On the other hand flair align
outputs nothing which is even more confusing. It would be helpful to have a summary at the end of flair align
about how many reads were uniquely mapped, multi-mapped, unaligned, etc.
Copy and paste the exact command you tried to run
First I tried running minimap2, followed by flair correct:
Then I tried flair align instead of minimap2 directly:
How did you install Flair?
conda create -n flair -c conda-forge -c bioconda flair
)What happened?
I attempted the first set of code on 12 different libraries using a job script, so all the code is identical. It ran as expected for 11 of the libraries. For this particular library (hb4) minimap2 produced 200k+ alignments as expected, but then flair correct quit after seconds with an empty output file and no error messages:
This is rather bizarre to me considering that it ran okay for all other 11 libraries with some output. I tried this multiple times to ensure it wasn't just a coincidence, and every time it produced an empty output file.
When I ran flair align first instead of minimap2, it finished and output a nearly empty BAM file (2774 bytes) that I can't read with samtools. The BED file is empty and there are no errors (see log output below). Now this really doesn't make sense. Why would minimap2 run fine, and flair align not run?
What else do we need to know?
What possible reasons could there be for the results I'm seeing? Obviously there must be some difference in this library compared to the others, but it can't be a detrimental difference because minimap2 was able to run an alignment without any problem. But flair align, which utilizes minimap, came back with nothing. And flair correct returned nothing when I provided it minimap's bam file.