search

alexdobin / STAR

RNA-seq aligner
MIT License
1.77k stars 495 forks source link

Possible to rerun STARsolo feature counting without alignment? #2132

Open jamestwebber opened 2 months ago

jamestwebber commented 2 months ago

I ran STARsolo with the --soloFeatures GeneFull, and I'd like to run it with GeneFull_Ex50pAS as well to get those statistics. Is it possible to post-process the aligned BAM file I already have? It seems like most of the work should be the same as before, but it wasn't clear to me if this is supported. Perhaps the BAM records have feature-specific metadata.

jamestwebber commented 2 months ago

I ended up re-running the whole thing with the chosen feature model, but it seems like the BAM files are very similar (perhaps up to some multi-mapping?). In the CellReads.stats files, the number of CB matches and total UMIs are the same, and the genomeU and genomeM columns are the same in the outputs.

So my impression is that like it might be possible to rerun the quantification from an existing BAM file, and this would be quite handy for trying out different models (and also for things like subsampling analysis). Does that seem like a reasonable thing to add?