is there any software for purity analysis for cnvkit

[worker000000]

Thanks a lot for updating such a powerful tool continously in the cnvkit docs, it recommend theta2, and give the command. but in fact the theta2 has someproblems and the author does not recommend using it anymore https://cnvkit.readthedocs.io/en/stable/heterogeneity.html#using-cnvkit-with-theta2

for tumor-only sample, there is a tool KhiabanianLab/All-FIT, but there seems to be noone knows how to prepare the input file, for paired samples, sequenza facets can calculate, but also calcaulate cnv, is there some recommended tools ?

[tetedange13]

Hi @worker000000 ,

As explained in this part from CNVkit documentation, several ways exist

• I guess your samples were not caracterized by a pathologist, to estimate "tumoral cellularity"?
• Tool you mentionned (All-FIT) uses "allele frequency" of several variants to do purity estimation, so you need to have results from variant detection analysis first (i.e. a VCF file)

Also you are talking about Theta2, but several other tools are mentionned in this part of CNVkit documentation => Have you tried each of them? => In my own experience PureCN works well. It requires to have normal samples, but regarding your other issue I guess you have some ?

If tools like sequenza and FACET can infer purity, you can use them? => Would be also interesting to compare CNV these other tools predict, with prediction made by CNVkit

Hope this helps. Have a nice day. Felix.