Project 2: Neutral tumor evolution testing

[tdelhomme]

Test neutral tumor evolution model described in Williams et al. on TCGA data.

Data: public MAF files

Script: R script, which output regression coefficient and slope of the model, by sample.

Guidelines: Loop on files

Project source code and documentation is hosted here.

[tdelhomme]

@NoemieL, you should have an R script computing this on a VCF file, you just should adapt this to MAF file. Be carefull, you have one MAF per cohort.
@aurelieGabriel would give a presentation on TCGA data and so also on MAF format.

[mfoll]

Note that Williams et al. selected only samples with a tumor purity >70%. Estimates of tumor purity for most TCGA cohorts are available in COSMIC under "ASCAT Ploidy and Purity Estimates".

[mfoll]

One difficulty is that you have to filter on allelic fraction (>0.12 & <0.25 for example), and MAF files don't always contain this information. VCF do, but are protected and are not filtered with the same QC. Maybe check in which cohort the allelic fraction is available in the MAF file with @aurelieGabriel?

[aurelieGabriel]

In every MAF file the following columns are now reported: t_depth (Read depth across the locus in tumor BAM) and t_alt_count (Read depth supporting the variant allele in tumor BAM).

[mfoll]

Good news! (It was not the case before).

[mfoll]

Tumor purity are also available in the biospecimen/slides tables available on TCGA data portal like here for LUAD for example

[mfoll]

To run a R script in your docker container:

• you need to make it work as a command line using Rscript.
• then you need to put it somewhere on GitHub
• when you create your Dockerfile, you will need to (you should use the biocontainer base image and then conda):
  • install R
  • install the R packages in R using RScript (see here for example)
  • download your R script and copy it somewhere in the docker image (see here for example)