SMARCB1 deletions in ATRT with current SEG to gene mapping

[jaclyn-taroni]

Purpose/implementation

In #182, @jharenza suggested we look at SMARCB1 deletions in ATRT. This analysis is crossed out on #186 because we were no longer looking to set LRR thresholds, but I was curious and thought it was worth looking at whether these show up given the SEG file to gene symbol mapping introduced in #195.

Issue

#186

Directions for reviewers

Rendered notebook is here.

Results

We don't pick up SMARCB1 deletions in ATRT and we don't capture SMARCB1 at all. This doesn't mean that the SEG file to gene symbol mapping is wrong, but it is worth being aware of. A possible update to this is to look at the genomic location of SMARCB1 in the package(s) we used for the mapping step.

Docker and continuous integration

Check all those that apply or remove this section if it is not applicable.

• 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.

[jaclyn-taroni]

@jashapiro figured out that the GenomicFeatures::genes step was filtering out SMARCB1 specifically because it was also mapped to some other alternative chromosome. We added filtering to chromosomes 1:22 with jaclyn-taroni@d085081. I've now run this notebook on the annotated_cn.tsv file that included that filtering step and these results look more consistent with what we would expect.

Here's the updated rendered notebook: https://jaclyn-taroni.github.io/openpbta-notebook-concept/seg-to-gene-mapping-atrt-chrom-filter.nb.html

cc @jharenza

[jaclyn-taroni]

I forgot to note: these results still are using CNVkit with no notion of ploidy, so the contents of annotated_cn.tsv will still probably change but the overall picture looks better.

[jharenza]

Yay!! Nice work @jashapiro and @jaclyn-taroni! One note: we should keep X&Y in if we can, as ATRX is on the X chr. I think, however, the sex chromosomes may have to be handled with caution. When we ran the CN tools, we used the germline_sex_estimate to correctly assign sex and therefore ploidy of those chromosomes, and I haven't looked into it, but for males, this may mean X&Y each show a loss of one copy as baseline and therefore. To do that, we can check the tools to assess ploidy for X/Y in males vs females.

[jaclyn-taroni]

I've made some updates now that #253 is filed. You can see the rendered notebook here: https://jaclyn-taroni.github.io/openpbta-notebook-concept/seg-to-gene-mapping-atrt-both-formats.nb.html

To my non-expert eye, this still looks good (cc @jharenza).

[jharenza]

Looks good, but are we missing 4 samples from ControlFreeC?

[jaclyn-taroni]

For the files I've used here, I chose to filter out neutral status so perhaps it could be that step?

https://github.com/jaclyn-taroni/OpenPBTA-analysis/blob/a5f8a6be13db24b3f45ed7e5f6d4b8736e11ed61/analyses/focal-cn-file-preparation/01-prepare-cn-file.R#L144

[jharenza]

Ahh ok. I will be interested to see the results of #128 to see if these 4 samples come back with Manta results. If not, I may be concerned that we may miss them. cc: @hongboxie and @fingerfen - something we can revisit once we get consensus calls.

[jaclyn-taroni]

3 samples are in CNVkit but not ControlFreeC and also ControlFreeC finds one gain which CNVkit marks as a loss (via the choices made over on #253): https://jaclyn-taroni.github.io/openpbta-notebook-concept/seg-to-gene-mapping-atrt-short-comparison.nb.html

[jaclyn-taroni]

This is now well covered via the PRs for #244