Enhancement Suggestion: GeneCounts with inputAlignmentsFromBAM

[Brunox13]

It would be very useful to be able to run --quantMode GeneCounts with a BAM file as an input. A typical scenario would be, for example, someone aligning their fastq files with STAR (--runMode alignReads), including --quantMode GeneCounts, then modifying the BAM file (e.g. filtering), and then trying to quantify the reads per gene again using the modified BAM file.

Currently, --runMode inputAlignmentsFromBAM (in order to be able to use BAM as input) is only possible with either --outWigType or --bamRemoveDuplicatesType (which, by the way, I only saw properly documented in the STAR manual 2.5.3a but not in the later/latest STAR manuals!), each of which produce their respective outputs but both of them completely ignore --quantMode GeneCounts.

I understand that this is currently best done using htseq-count (whose results are mostly identical to those with STAR's --quantMode GeneCounts) but using the same software (STAR) for alignment and comparing the gene counts before & after the BAM file was modified would be ideal in terms of smoothness of workflow and scientific reproducibility (version tracking when following someone else's workflow etc.). It also seems that all the necessary functionality is already there, so it might be an easy feature to implement.

Thank you!

[alexdobin]

Hi Bruno,

this feature has been on my list for a long time.
It's not as easy as it sounds, as it would require careful checking of the BAM file formatting.

I will add the description of the --runMode inputAlignmentsFromBAM option in the manual, not sure why it disappeared.

Thanks!
Alex