AlexsLemonade / alsf-scpca

Management and analysis tools for ALSF Single-cell Pediatric Cancer Atlas data.
BSD 3-Clause "New" or "Revised" License
0 stars 1 forks source link

Testing quantification of spatial libraries using --sketch alignment with Alevin-fry #163

Closed allyhawkins closed 2 years ago

allyhawkins commented 2 years ago

Following up on https://github.com/AlexsLemonade/alsf-scpca/pull/162#issuecomment-993945130, I went ahead and ran the two samples we have been using with --sketch mode for alignment, but keeping the --unfiltered mode as that retains the most similarity with Spaceranger. I then added these runs to the existing notebook (in order to not repeat a lot of the same code) and am showing the results grouped by tool + configuration.

The main output that I was interested in here was checking to see if use of --sketch would increase the total counts and number of genes detected/cell observed when quantifying with Alevin-fry and that is in fact the case. The distributions look much more similar to Spaceranger and the plots showing the spots overlaying the tissue look almost identical to Spaceranger. I kept the addition of --sketch pretty simple, as it looks to mostly duplicate the results seen with Spaceranger. However you do still see a small subset of genes that shows higher gene expression in Spaceranger, like we were seeing in selective alignment, but to a lesser degree.

This PR is stacked on #162.

The updated html with the new plots can be found here.