Enable gc correction by default in somatic cnv panel workflow

[mwalker174]

• Substantially improves results on FFPE samples

[codecov]

Codecov Report

❗ No coverage uploaded for pull request base (master@1a290ae). Click here to learn what that means.
The diff coverage is n/a.

@@            Coverage Diff             @@
##             master     #5966   +/-   ##
==========================================
  Coverage          ?   86.934%           
  Complexity        ?     32787           
==========================================
  Files             ?      2014           
  Lines             ?    151468           
  Branches          ?     16642           
==========================================
  Hits              ?    131677           
  Misses            ?     13729           
  Partials          ?      6062

[samuelklee]

Thanks for making this change! I might include more detail with the note ("Substantially improves results on FFPE samples"), for posterity---it's probably true, but we can't really say anything definitive with only N=1 and without the cross-validation procedure I mentioned on Slack. That is, the higher degree of denoising might just be an artifact of effectively removing more PCs with GC-bias correction (since I'm assuming the same number of PCs were explicitly removed in both cases), but it's possible that removing the optimal number of PCs without GC-bias correction could achieve a better result. Since our correction procedure is relatively naive, there may also be some dependence on bin size. However, I think it's probably not worth a detailed analysis, and that it's generally safe to enable correction by default.

[samuelklee]

@mwalker174 Here's that branch. Let's just go ahead and enable this on the gCNV side as well?