interpretation of CNVKIT plot

Question

[chenglei1865]

Hi, I'm a new to CNVKIT, I tried to use CNVKIT to analyze my WES data. Finally, I got a plot, as following:

CNKN2A homozygous deletion frequently occurred in some cancers. In my plot, did CNKN2A gene (red arrow denoted) habor homozygous deletion? how to tetermine wheter a gene habor homozygous deletion OR LOH via CNVKIT scatter plot? Hope anyone give some adivce! Appreciate!

Answer 1

[jared.andrews07]

There is no hard and fast threshold to use for this. CNVkit's documentation on the call command and calling losses and gains may be helpful.

In short, you kinda have to eyeball it based on previous knowledge and these plots, or if you have an idea of tumor purity, you can let the call command try to re-scale the scores for each segment and round to the nearest whole number based on that. If you have variant information, that's an additional level of information that can be incorporated, as details in the call command docs.

[chenglei1865]

Thanks very much for your reply. I wonder whether homozygous deletion could be inferred via copy number, for instance, whether a copy number <1 indicate homozygous deletion. In my case, the log2 ration is -1.293 and I converted it into actual copy number, copy number is 0.816, so, whether this copy number indicated homozygous deletion? Is my understanding correct?

[jared.andrews07]

Likely so, in that case, but I'd really recommend reading the links above. CNVkit's call command will spit out absolute copy numbers. You can manually define the log2 ratio thresholds used if you'd like.