Planned Analysis: Immune/Stroma characterization across PBTA

[PichaiRaman]

Determine immune/stromal composition using software such as xCell and correlate back to genetic and phenotypic characteristics.

[jharenza]

A new method, MIXTURE, seems to do a bit better for tumor microenvironment cell deconvolution than previous methods and was validated with flow cytometry. Code here: https://github.com/elmerfer/MIXTURE.App

[cgreene]

I read the MIXTURE paper but didn't find the supplement. The supplement seems like it's pretty necessary to understand how well we'd be able to apply this to our own data. I am excited about the potential of the method, but also a bit cautious after reading the details.

I did tweet about the supplement. bioRxiv keeps moving things around on the website, so it's possible I missed it.
https://twitter.com/GreeneScientist/status/1159799316984356866?s=20

[jharenza]

Oh good catch - I don't see it either.

[cgreene]

The author ( @elmerfer ) provided a link to the supplement in a comment on bioRxiv
https://www.biorxiv.org/content/10.1101/726562v1#comment-4575975402

I haven't gotten to read it yet but it would be good to go through it.

[komalsrathi]

I would like to take this on, potentially using MIXTURE as well as immunedeconv which incorporates methods like CIBERSORT and xCell along with others to estimate immune cell fractions from bulk RNA.

[jharenza]

@komalsrathi great! Looking forward to seeing your analysis!

[jaclyn-taroni]

Hi @komalsrathi, do you have an estimate of when we can expect the first pull request adding this analysis? This will help us factor in review as we plan our workload. Thank you!

[komalsrathi]

@jaclyn-taroni Potentially by next Friday? (it could be earlier but just giving myself some extra time). Does that sound reasonable? Let me know.

[jaclyn-taroni]

Yes sounds good - thank you!

[komalsrathi]

@jaclyn-taroni @jharenza It looks like I will be able to create a branch and push my existing scripts today but for the actual pull request, it will take a couple days more - cibersort needed me to request access to the source scripts which I just got my hands on. I would like to finish a correlation plot between all methods used and another cell type visualization per histology (boxplot) before I submit the pull request. Let me know what you think.

[jaclyn-taroni]

@komalsrathi you can create a pull request for what you have so far and then add the CIBERSORT steps, correlation, and additional plot in a later pull request(s). Generally pull requests with fewer parts are easier to review. Would that work on your end?

[komalsrathi]

I think that should be doable - let me see if I can add a branch with at least 1-2 clean scripts. Thanks!

[jaclyn-taroni]

Closing all planned analysis tickets in favor of opening new proposed analysis/updated analysis tickets as needed.