Explore Variant_Classification's Impact on TMB: Part 2: Definition discrepancies - how often do they come up?

[cansavvy]

Purpose/implementation Section

⚠️ This PR includes changes filed in #739; that should be reviewed first

What scientific question is your analysis addressing?

This address part 2 of #729

2. How often do these: Translation_Start_Site, Nonstop_Mutation or Splice_Site come up? (These are not included in the FoCR definitions, but are included in maftools definition.

There's not very many mutations that are affected and given the results of part 1 (#739) I do not consider this to be something that should be a big concern moving forward.

What was your approach?

Pretty straightforward:

• Import consensus mutation files for TCGA and PBTA.
• Make barplots of the Variant_Classification variable.
• Label which are included in the FoCR nonsynonymous definition and which are only noted by maftools nonsynonymous definition.

I also added an option to calculate_tmb.R script that allows us to choose which nonsynonymous mutation definition we want to use for filtering. So now there's --nonsynfilter_maf or --nonsynfilter_focr.

What GitHub issue does your pull request address?

#729

Directions for reviewers. Tell potential reviewers what kind of feedback you are soliciting.

• Note that this uses the consensus file which is based on strelka, mutect, and Lancet and does NOT have coding region filtering done. So it doesn't exactly answer about TMB directly, but is related.
• Does this address the scientific questions at hand?
• Are there further analyses we'd like to see here?

Is the analysis in a mature enough form that the resulting figure(s) and/or table(s) are ready for review?

Yes

Results

What types of results are included (e.g., table, figure)?

Here's the rendered notebook:
explore_var_class_discrepancies.nb.html.zip

Reproducibility Checklist

• The dependencies required to run the code in this pull request have been added to the project Dockerfile.
• This analysis has been added to continuous integration.

Documentation Checklist

I haven't added a README since this is a side analysis. The main information is in the Rmd. It's not a lone standing module so I also didn't add it to the table. But if we would like these items to be added, I can.

• This analysis module has a README and it is up to date.
• This analysis is recorded in the table in analyses/README.md and the entry is up to date.
• The analytical code is documented and contains comments.

[jashapiro]

I like this addition!

[jashapiro]

Now that I have looked at this NB, I will confirm what I think I said in my review of #741, namely that I think these two notebooks can be combined, as this is kind of the crux of things at this point.

If we start with the MAF file, we should be getting the same mutation counts by either script: any discrepancy (even off by 1!) indicates a difference in methodology and/or an error in one script or the other. At this point, we know that the differences are small in counts, and the question is whether the FOCR vs. maftools method is accounting for that difference. I suspect it is, though there may be some remaining edge cases that could result in some remaining differences.

The second part (covered more in #741) is a difference in the definitions of the reference regions. I just looked back at that data and have another comment there...

[jashapiro]

Why are we not starting with the full maf file in the data directory here? Using these sub files seems like it could be a source of inconsistency.

[cansavvy]

I switched this for PBTA, but realized the TCGA tmb data is not in the data release, so I'll file an issue about that.

[cansavvy]

#732 (comment)

[cansavvy]

I'm not exactly sure if this addresses all your suggestions, but I think it does? I like keeping the notebook here and the one in #741 separate, since they are slightly different questions (even though they are related).

[jashapiro]

After our discussion, the comparison of focr and maftools results will go in #741, so this should be fine as is!