How are the log2 values in CNVkit .cnr files calculated?

Question

[Jordan]

Hi,

I'm relatively new to CNV analysis. I ran cnvkit on WGS samples (4 tumor/normals) and the normals were pooled.

I don't quite understand how the log2 values in .cnr files for CNVkit are calculated.

Here is a result from one of my sample:

normal.cnn:

chr1 11129 12259 DDX11L1 28.2027 4.81772

tumor.cnn:

chr1 11129 12259 DDX11L1 373.432 8.54083

tumor.cnr:

chr1 11129 12259 DDX11L1 373.432 -0.23305 0.805129

As can be seen from .cnn files, the tumor has higher depth, but in .cnr file it has negative log ratio. I looked at other normal samples in the same region and they almost have similar depth and log values in their .cnn file.

So, why is it the log ratio .cnr file for this bin negative?

Thanks for the help!

[Arup Ghosh]

check this reference: https://cnvkit.readthedocs.io/en/stable/fileformats.html#bin-level-log2-ratios-cnr

[Jordan]

I read that part of the manual, but I’m still not clear about it. If I’m understanding that document correctly, weights are used for log2 ratio calculation?

[linouhao]

the doc is very unclear, the fix command used the reference, which used the normal and tumor sample data, and cnr just uses the cnn file from tumor and the reference, it filter some region, but the calculation is very unclear

Answer 1

[Kevin Blighe]

Standard log2 ratio is reported as log2; 'weighted' ratio is reported as weight

Kevin