Closed Fponelle closed 2 years ago
Is this happening with CNVkit version 0.9.8?
I'm sorry I forgot to specify the version but yes I am using 0.9.8
Hi @etal , I would like to contact you again to find out if you have found some answers to my problem?
Hi @Fponelle, looking at the code, it looks like the log2 values in the bintest output are residuals: https://github.com/etal/cnvkit/blob/25d91cdb962639c98d041c3b05c7de659c41dc70/cnvlib/bintest.py#L26
If the segments are provided, the residuals are calculated compared to the segment to which the bin belongs, and otherwise to the whole chromosome: https://github.com/etal/cnvkit/blob/25d91cdb962639c98d041c3b05c7de659c41dc70/cnvlib/cnary.py#L417-L429
I suspect that you're running bintest with the segments specified. And so, given that the whole BRCA2 is duplicated, the change of CN of any given bin is not relevant compared to the BRCA2 segment.
Could you try re-running without the -s
parameter to see if this fixes your issue?
In any case, it looks like documentation on bintest could be improved to reflect this behaviour.
Hi @tskir Thank you for your answer, my problem is solved. By default in the "batch" command the segments are specified, you must launch the "bintest" command separately to be able to remove this parameter.
Yes, that's right. With -s, you're telling bintest that those are the alterations you already know about, and asking for any additional bin-level alterations. Maybe that subtraction should be used only for the reported p-values, and not alter the log2 values in the output.
Note to self — Leaving this issue open to improve bintest documentation in the future
Help wanted: for someone to document bintest and add the section to the relevant part of CNVkit docs
Hi,
Why log2 values are different between cnr file and bintest.cns file?
Here is my example:
- In .cnr file, all bins on BRCA2 have a log2 between 1 and 2.5 suggesting and amplification of the entire gene:
- .cns file agrees with .cnr file showing a log2 for the whole gene of 1.64:
- call.cns reports a CNV on the BRCA2 gene by counting 7 copies with a mean log2 of 1.6:
- But in bintest.cns file, for the same bins, log2 are really different, some of them are negative: (Initialy those bins were not reported in bintest file, I modified alpha value to 1 in bintest.py to understand why my bins were not reported as significative.)
Can you help me understand why there is such a difference?
Thanks
Flora