Quantification comparison, PR2 of 2: Plot mapper comparisons

[jashapiro]

This PR adds a notebook with comparisons of various mapper options, looking at concordance with the "standard" of Cell Ranger.

This PR hopefully completes #9 and #2 (at least for now), working from the backbone of #35

⚠️Note: the target for this PR is #39 , but that should update to master once #39 is merged... that is part of what I am testing here: properly stacked PRs!

The easiest way to view the rendered notebooks is with htmlpreview: https://htmlpreview.github.io/?https://github.com/AlexsLemonade/alsf-scpca/blob/jashapiro/compare-quant-2-analysis/analysis/quantifier-comparisons/02-compare_quants.nb.html

Results summary

I am here copying with slight modification the comments that I left in #35:

The top line result is that the results from Alevin are much more in line with the Cell Ranger baseline than Kallisto's results. There is a subtle further improvement with the addition of decoys, but it is small.

As discussed earlier (#24 (comment)) Kallisto seems to consistently give somewhat higher expression values for lowly expressed genes; Alevin does this too relative to Cell Ranger (to a lesser extent), which leads me to suspect that this is a result of pseudomapping, which may be partially offset by Alevin's selective alignment. Cell Ranger may also be underestimating some genes if freagments are multimapped and discarded.

Note: while this analysis was done using cDNA transcripts only, I have no reason to suspect (partly based on results from #24) that using the full transcriptome would change the results at all. Indeed, #22 suggests that the full transcriptome is better, but I only ran Cellranger with cDNA, so I was limited to that comparison for this notebook.

Questions for Reviewers

Are the analyses presented here reasonable? Are there any glaring errors of logic or omission?

If there are new analyses you might suggest, we may want to add those as a separate issue for future analysis.

[jaclyn-taroni]

Going to return my "this could be easier to read" review first.

[jaclyn-taroni]

LGTM other than minor things! (And personal preferences, which for the most part I have refrained from leaving here – I prefer lots of inline comments in the code blocks because I think it makes life better for six-months-from-now-me.)

[jaclyn-taroni]

Did you want to add a : and something more descriptive like you have for the earlier notebook?