Somatic copy ratio alterations in benign case samples
Hello I have been following tutorial(s) 11682 and 11683 for sensitive detection of copy ratio alterations in benign tissue. Previously we have used other tools, such as Mutect2, to look at CNV when we had matched normal germline samples. Now however we would like to use a panel of normals, consisting of 47 samples, outlined in the tutorial without matched normals to look for copy ratio alterations in our benign case samples.
I have run several case samples through the tutorial and noticed that after running ModelSegments command some of the sample plots looked similar to those seen in figure 8D of the tutorial while others had many more segments and more noisy. I am not sure if this has to do with the denoising steps or the ModelSegments step? Any ideas and suggestions to address the discrepancy would be a great help! Thank you
Can you please post the sample plots.
No problem! I will only post two of the 20 otherwise my computer freezes up for some reason. Only 6 of the 20 look like sample #160. Thank you for your help!
- The oversegmented sample looks pretty noisy and has a lot of bins with coverage dropout (i.e., the points at zero copy ratio), which makes me think the PON samples might not be representative of the cases (perhaps they might even be sequenced with different target kits). This will result in poor denoising.
- Can you please provide more info about how you picked your PON and
- Also take a look at the documentation in the tutorials about underlying assumptions of PCA denoising.
- In some cases, poor denoising can be mitigated by selecting more conservative segmentation parameters on a per-sample basis (or just dropping those samples from the analysis outright).
Note: Sometimes it is difficult to apply the general QC recommendations when the data is very noisy. In those cases it is important to understand the underlying assumptions of PCA denoising.
I apologize as I did not submit my response last week. To answer your points:
1) All samples, case and PON, were sequenced using the same target kit and methods.
2) The samples in the PON were previously matched germline for cases and controls taken for an earlier analysis using a different pipeline. They are saliva samples taken at a later date than the tissue sample but processed as stated in point 1.
3/4) I will take some time to look through the documentation regarding making assumptions about the PCA analysis.
Thank you for your help!
Please sign in to leave a comment.