Tumor fraction and CNV: a loop?

[mrfox]

I am trying to detect somatic CNV using matched tumor/normal exome sequencing data.

I found that popular tools such as EXCAVATOR, ExomeCNV etc require tumor fraction or cellularity as an input parameter.

At the same time, tumor fraction estimation tools such as THetA require CNV result as input.

Could anyone recommend a simple tool that could estimate tumor fraction without using CNV data, or a CNV detection tool with built-in function to estimate tumor fraction?

[stefanoberri]

Hi.

It is partially a loop. We provided a solution with CNAnorm a tool described in this paper

http://www.ncbi.nlm.nih.gov/pubmed?term=22039209

but we awknoledge that there is not a single solution to the problem

Imagine your tumour is 67% pure, triploid for chr1 (2/3 of cells have the amplification) and diploid everywhere else.

This will give you the very same data when the tumour is 50% pure, esaploid for chr1 and tetraploid everywhere else.

If you have more complex pattern of amplifications and BAllele data, your model could still come up to a decent guess. If you have heterogeneity the problem becomes more difficult.

However, if you know the various copy number states, detecting tumour content (and even heterogeneity) becomes easy or at least possible. Similarly, if you have tumour content, detecting copy number states accurately becomes possible.