cnvpytor.utils - WARNING - Problem with fit: Runtime Error. Using mean and std instead fitting parameters!

[kaqisekuzi]

          Hi @suvakov 

I sent the log file to your email. Below is the Manhattan plot of the two samples. Does this Manhattan plot have anything to do with the warning message?

warning message: cnvpytor.utils - WARNING - Problem with fit: Runtime Error. Using mean and std instead fitting parameters!

log file: sample_3964_And_sample_4352.log.txt

Originally posted by @kaqisekuzi in https://github.com/abyzovlab/CNVpytor/issues/171#issuecomment-1491467672

[kaqisekuzi]

Hi @suvakov @abyzov Does this Manhattan plot have anything to do with the warning message? warning message: cnvpytor.utils - WARNING - Problem with fit: Runtime Error. Using mean and std instead fitting parameters!

[suvakov]

Warning message is expected due to shallow coverage (around 1x according to log files) and it is not related to Manhattan plot. Is this cancer sample (first plot)?

Thanks, Milovan

[kaqisekuzi]

@suvakov Yes , first plot is a cancer sample. Do these warnings have any effect on my results? Is it possible to ignore these warning messages?

Thanks, kaqisekuzi

[suvakov]

You can safely ignore warning messages. In the first example, it's unclear what copy number 2 level refers to. Normalization is typically performed based on the mean level of autosomes, but in cancer cases, this may differ from the actual copy number 2. Our 2D caller addresses this issue by using only diploid regions for normalization based on B-allele frequency (BAF). However, this approach may not be applicable here due to the low coverage. My suggestion is to manually normalize the levels by assuming that the beginning of chromosome 1 (or any other region where you believe it is most likely to have 2 copies) has a copy number of 2.