Roth-Lab/pyclone-vi

CCF estimation for minor_cn=0

Opened this issue · 1 comments

Hi,

I am running Pyclone VI for my WGS data. The CCF doesn't seem right for mutations with minor_cn = 0. Here's an example from my dataset. I am interested in one cluster output from Pyclone VI. I have put the mutations in this cluster into 5 groups by their (major_cn, minor_cn) values. Pyclone VI estimated the CCF for this cluster as 0.93. Given this CCF and the tumor purity of this sample (0.55), I calculated the VAF (blue bar) for each group and compared to the VAF directly estimated from ref and alt read count (y-axis). Please see the plot below. Do you have any suggestions for why this might be happening or how to fix the issue? Thanks!

image

PyClone-VI allows the number of mutant alleles to be 1 to major_cn. By contrast you are assuming the number of mutant alleles equals major_cn in your calculation of VAF. So PyClone-VI is inferring there is 1 mutant and 1 reference allele in the 2,0 mutations. This would imply that the copy number change happened before the mutation.

The current behaviour is equivalent to the major_cn prior from the original PyClone software. The calculation you are doing is based on the total_cn prior.