Hello! I'm performing a single-cell genome-wide copy-number profiling by low-pass sequencing (coverage: 1X, 0.5X, 0.1X). Would anybody suggest me how to determine which is the optimal bin size in the genome segmentation step? Usually it's set to 0.5M or 1M when the genome coverage is as low as 1X. I would like to know which is the best one for my coverage, any advice? Thank you!
•
written
by
am •
The first author of this paper has developed an interactive app that lets you play around with your sample characteristics and see how that would affect the required read depth to be able to perform CNV calling: https://gmacintyre.shinyapps.io/sWGS_power/
Bizarrely, they do not seem to have done any adjustment for GC content, which will affect coverage at different genomic regions. As such, on face value, this appears to be yet one more copy number program that is not addressing all of the issues.
This is not a CNV program, just a tool to help you estimate the depth you need based on sample characteristics before you do your sequencing.
Gracias por la clarificación / Thanks for confirming.
The answer was not really for you, but rather so that other people don't just dismiss the tool because of your comment. I think it can indeed be useful.
Fantastic Spanish, by the way.
De nada
Log in to answer this question.
Which tool are you using for the analysis?
I'm testing different tools such as CNVkit, Gingko, Control-FREEC... In the CNVkit manual for example I read: "Increase the “target” average bin size (--target-avg-size), e.g. to at least 1000 bases for 30x coverage, or proportionally more for lower-coverage sequencing". I would like to expand this concept.
I am not sure about the others, but Control-FREEC can automatically determine the optimal bin size.
I second the use of Control-FREEC. It is a truly great program suite, that can determine copy number from WGS, WES, or targeted seq.
Thank you Igor and Kevin. Yes, the coefficientOfVariation parameter in Control-FREEC allows to automatically determine the correct window size.