How do you know the best segments in cnvkit after filtering?

Question

[Jordan]

Hi,

I'm working on WGS data using cnvkit. I filtered the segments using segmetrics and call. Here are my steps:

cnvkit.py segmetrics tumor.cnr -s tumor.cns --ci --sem -o tumor.segmetrics.cns
cnvkitpy call tumor.segmetrics.cns --filter ci -o tumor.segmetrics.call.cns

I have about 300 segments at the end of this step. Is there a way to tell, which are statistically the best segments?

For e.g., two rows in tumor.segmetrics.call.cns file looks like this.

chromosome      start   end     gene    log2    cn      depth   probes  weight
chr1    125086529   125146265   -   0.0648392   2   143.179 52  44.4282
chr1    125146265   125175570   -   -1.35037    0   71.7419 24  14.5482

Is there a way to tell which segment you trust more? I understand that, more the weight better the reliability.

But I was wondering how you would normally pick the best segments?

Thanks in advance!

Answer 1

[brunobsouzaa]

What helps me on my analysis is: - get only cn values that are not 2; - output the ref depth for each segment on the spreadsheet and the depth ratio. - correlate the cn values with depth ratio and weight!

Also you can build your own database with the most frequent calls on different samples!

Hope it helps.

[Jordan]

Thank you! Looking at the weight and CN values looks like the best way to go.

[brunobsouzaa]

Try adding another CNV caller to your analysis and choose common results!