Runtime questions
bremsen opened this issue · comments
Hello,
Thanks for this great tool!
-
How does the run-time scale with number of CNVs?
I understand that CNVs are converted to pseudo-SSMs, so do CNVs effectively increase the number of SSMs? -
Can I use PhyloWGS on CNVs alone? I know we can do SSMs alone with no CNVs, but what about the other way?
-
Is there subsampling for CNVs? I see a flag for writing non-sampling CNVs to an output but I don't see a flag for number of keeps to keep.
Thanks!
Hi!
- Yes, each CNV is equivalent to adding an additional SNV. In addition, each SNV affected by a CNV requires a slower computation than an SNV unaffected by CNVs.
- I don't think there's any technical restriction on this, but I've never tested this.
- How many CNVs do you have? We've never used data with thousands of CNVs, so I don't think we've ever implemented a subsampling for CNVs. You can always manually subsample rows in the resulting processed CNV file
Hi Amit,
Thank you for the responses.
I have ~4000 CNVs which were inferred using TitanCNA. I didn't do any filtering on the CNVs. Are CNVs that do not overlap with SNVs used at all? As I understand, CNVs are used to correct the VAF of each SNV.
Also, Is there a way to get a list of SNVs and CNVs in each inferred subclone? The tree viewer tells us how many SNVs and CNVs but not which ones specifically.
Thank you!
Hi!
You can still use CNVs that don't overlap any SNVs to make a more complete genotype for each cluster, because CNVs are clustered just like SNVs. If you're only interested in the SNVs they're probably not needed.
Yes, the identity of the mutations assigned to each cluster in each tree are found in the trees.json file.
Hi Amit,
Thank you for your responses. I have yet another question.
Do you know why the parser groups many CNV segments under a single CNV ID? I see one CNV ID (c0) has over a thousand CNVs under "physical CNVs".
Thank you.
@jwintersinger Is this a clonal CNV parsing thing?
