search

broadinstitute / PhylogicNDT

Other
70 stars 39 forks source link

CCF of child cluster greater than parents #52

Open nadia35 opened 3 years ago

nadia35 commented 3 years ago

Hello!

We're testing PhylogicNDT on multi-sample sequencing data, and looking to better understand cluster CCF. In this run (see below), cluster 2 is descendent from cluster 1. However, samples T0, T6, and T8 show a higher CCF of cluster 2 mutations than cluster 1, and their confidence intervals do not overlap. Are we using the tool correctly, and is there a way to ensure that cluster CCFs follow the sum rule?

Thank you!

clustering_plot tree

jcha40 commented 3 years ago

Hello,

There are a couple of possibilities: One is that some of these clustered mutations are artifacts, in which case they should be blacklisted. Another is that the CCF calculations for some samples were based on inaccurate purity/ploidy calls.