Classifying copy number alterations

Q&A
1

Hi Everyone,

The documentation states that the gene-level copy number values in the DepMap download are log2(copy_ratio + 1). Presumably copy_ratio = copy_number/2, in which case it is easy to compute quantitative gene-level copy number values. Can anyone provide guidance on thresholds for classifying gains and losses? For example, the GISTIC pipeline produced both quantitative and discrete copy number calls (-2 = homozygous deletion, -1 = heterozygous deletion, 0 = copy neutral, 1 = low gain, 2 = high gain) for the various TCGA cohorts. The general consensus is that -0.1 and 0.1 were thresholds for classifying gains and losses, respectively, based on the quantitative gene-level copy number values from GISTIC. However, these values went through an extensive normalization procedure, so I don’t think there’s no reason to expect the same thresholds to be applicable to the CCLE data. Any thoughts would be greatly appreciated.

2

Hello,

We currently don’t have a specific threshold recommendation, but perhaps the discussion in this thread could be helpful.

And in case you are interested in absolute copy number calls instead of relative to classify CNAs, we currently have data generated by the ABSOLUTE algorithm on the CCLE lines. You can find the file called CCLE_ABSOLUTE_combined_20181227 in the “CCLE 2019” data set on the download page. We are also working on running PureCN on all of our current DepMap data for absolute copy number calls, and we’re hoping to include it as part of the release in the future.

Hope this helps!
Simone

3

Dear simz,

could you please explain what Modal_HSCN_1, Modal_HSCN_2 and LOH mean? I struggle to understand what “major” and “minor” means here. Is LOH a deletion of one copy of the segment?

Also, is ploidy in ABSOLUTE - absolute or is it a divisible of these numbers?
Thank you!