CNV calls data release 2

[fpbarthel]

To-do list of items related to finalizing CNV callset for data release 2.

• [x] Optimize HET calling to call more hets per sample. See: https://github.com/TheJacksonLaboratory/GLASS/issues/103
• [x] X-chromosome hets in males? TITAN results in chrX in males look bad. Can we improve this? Probably due to no/few "hets" being called here. We are moving away from TITAN so no longer relevant
• [x] Can we improve calling homozygous deletions? See my post on the GATK forums. This was done and related to the first point, part of GATK4 update. Re-wrote pipeline to correct for this
• [x] Re-run CNV calls with larger number of patients with known sex (current batch has no sex information for entire Columbia cohort, annotation for this was recently updated)
• [x] Perform CNV QC on the GATK denoised copy ratios and modeled segments and re-populate CNV blocklist. @EmreKocakavuk @Kcjohnson
  • [x] Independent QC of all GATK results @Kcjohnson @EmreKocakavuk (tiny contribution by @fpbarthel )
  • [x] Integrate independent QC results @Kcjohnson
  • [x] Final pass GATK CNV QC after the Sequenza results have been QC'ed. Minor adjustments might need to be made @Kcjohnson @EmreKocakavuk
• [x] Run Sequenza to get ploidy/purity values. @EmreKocakavuk @fpbarthel
• [x] Run TITAN @fpbarthel
• [x] QC Sequenza results. @EmreKocakavuk @Kcjohnson
  • [x] Compare segmentation and purity/ploidy to @wisekh6 sequenza
  • [x] Compare segmentation and purity/ploidy to @anderkj2 sequenza
  • [x] Compare ploidy/purity values to TITAN values, @Kcjohnson can point to these
  • [x] Compare segmentation to GATK segmentation. Inspect for hyper segmentation as an indicator for poor quality.
  • [x] Determine whether there are viable alternative solutions for Sequenza. Examine values that have similar -log posterior probabilities.
• [x] Use Aneuploidy value metric to scan for 0.0 samples that indicate no CNVs. These should be re-evaluated for blocklist.

[fpbarthel]

@moskalika @Kcjohnson please comment on completeness of sex information for new dataset as well as any (pending) updates to the old data

[moskalika]

@fpbarthel we have all genders except for 2 Dana-Farber cases; we are waiting for clinical data from them.

I also made a few changes to the clinical.surgeries table (consolidating one patient; adding TMZ info for 7 surgeries which are now '0'; initial diagnosis histo/grade/surg_interval)

[Kcjohnson]

We do not currently have any patient sex information from the DKFZ cohort. We may receive this information at a later point, but there's no guarantee. If chrX and chrY were not removed from the sequencing data, we may be able to reliably infer this information.

[fpbarthel]

@moskalika just to confirm that:

• these (non-DK) are currently missing sex?
• they will not be updated with this info anytime soon, except for the two DF cases (do we have an ETA for those?)

GLSS-DF-0013
GLSS-DF-0014
GLSS-MD-0085
GLSS-MD-0119
GLSS-MD-0120
GLSS-MD-0126
GLSS-MD-0135
GLSS-MD-0138
GLSS-MD-0140
GLSS-MD-0141

[moskalika]

@fpbarthel we are waiting for clinical data for the two DF cases. The MD Anderson cases from the first data freeze will not have genders.

[fpbarthel]

CNV data is now complete, closing issue