tetedange13
commented
3 years ago Hi @ashbig,
Not an author of CNVkit, but indeed it looks like a segmentation failure To be sure I well understood your case:
- You are expecting an amplification of whole MYC gene? (I mean not of a single or several exons, whole gene should be amplified)
- Your gene seems well amplified, regarding your ".cnr" only
- However, no matter segmenting method used, your amplified portion is always included into a larger segment (and so that predicted segment-level log2 value is not what you should obtain)
Would it be possible for you to share chr8 portion of your ".cnr" ? Or at least complete portion of segment predicted by CNVkit ? (i.e.: "chr8: 127331517 - 128114987" and not just MYC portion)
As a hot fix, have you tried using cnvkit.py bintest? (aimed to call more focal bin-level CNV) ?
=> Try it with and without ".cns" file produced (cuz results change if ".cns" is provided or not)
=> Not sure about that, but we never know ?
Hope this helps. Have a nice day. Felix.
Seeking advice on calling small clinically relevant copy number events.
1 ddPCR confirmed cell line sample for 8 copies of MYC was run using the batch command (hybrid capture). The output .cnr file shows the correct relevant copy number change.
chromosome start end gene depth log2 weight chr8 127736518 127736698 MYC 3062.61 1.76817 0.903265 chr8 127738213 127738493 MYC 5458.27 1.63719 0.930098 chr8 127738493 127738773 MYC 5959.46 1.60974 0.935167 chr8 127738773 127739053 MYC 5586.31 1.61023 0.9426 chr8 127740377 127740677 MYC 4535.27 1.76197 0.942491 chr8 127740677 127740977 MYC 4552.02 1.7254 0.933897
I have tried all of the available segmentation methods in the CNVkit 0.9.9 library, the most promising were hmm methods pasted below, but this number is still far below the average seen in the .cnr file which appears to be ~7 copies and fairly accurate.
chromosome start end gene log2 depth probes weight chr8 127331517 128114987 MYC 0.940634 2204.08 14 13.1454
Does anyone know how I might be able to capture this change? I know this is a small region at ~4kb, but with such steady values I feel even CBS should detect this and I have indeed seen a similarly small gene (MYCN) accurately segmented with even fewer probes.
Any advice/troubleshooting tips are appreciated!