search

amf71 / ParallelGDDetect

A package to detect distinct genome doubling events which have occurred in different samples from the same tumour
3 stars 2 forks source link

Output of WGD in individual gd_clusters #3

Open Nikki-Burdett opened 1 year ago

Nikki-Burdett commented 1 year ago

Thanks very much for this great package When reading your paper and the expected output I interpreted that ParallelGDDetect would define which clusters had clonal/subclonal WGD (and possibly also how many GDs) and which do not, but after running the tool on 10 patients (with multi-samples), I get a column titled 'gd_clusters' which is blank (not NA). Have I misinterpreted, has something failed, or is there not enough confidence for the tool to have called this?

amf71 commented 1 year ago

Dear Nikki-Burdett,

Unfortunately its often very difficult to assign the GD events to clones. Sometimes its possible and is something we've worked on a little since this publication, but this version of the tool doesn't attempt to do this formally - only to assign the GD events to particular tumour samples. If you do have clusters in the 'gd_clusters' column the most terminal of these on the phylogenetic tree may be the GD clone but sounds like you don't have any of these. This could mean a few things a) there are no subclonal GDs in your dataset, b) Any subclonal GDs occured very soon after the MRCA and hence no/very few subclonal mutation were doubled and the subclonal GD can only be inferred from the pliody which the tool should be doing, c) something about the input could be incorrect. The mutation copy number can be difficult to get right and is dependant on your purity/ploidy solution. You should find a multimodal distribution of mutation copy number with modes at 1 and 2 (possibly also three).

Hope that's helpful!

Alex

Nikki-Burdett commented 1 year ago

Thanks Alex. I do have some with subclonal WGD, but helpful to know that the tool isn't formally doing this at this point. It would be a great feature if successful in future! I think if I want to assign WGD to clusters I will have to think about this in another way Thanks very much for your reply

On Wed, Jun 14, 2023 at 1:30 AM amf71 @.***> wrote:

Dear Nikki-Burdett,

Unfortunately its often very difficult to assign the GD events to clones. Sometimes its possible and is something we've worked on a little since this publication, but this version of the tool doesn't attempt to do this formally - only to assign the GD events to particular tumour samples. If you do have clusters in the 'gd_clusters' column the most terminal of these on the phylogenetic tree may be the GD clone but sounds like you don't have any of these. This could mean a few things a) there are no subclonal GDs in your dataset, b) Any subclonal GDs occured very soon after the MRCA and hence no/very few subclonal mutation were doubled and the subclonal GD can only be inferred from the pliody which the tool should be doing, c) something about the input could be incorrect. The mutation copy number can be difficult to get right and is dependant on your purity/ploidy solution. You should find a multimodal distribution of mutation copy number with modes at 1 and 2 (possibly also three).

Hope that's helpful!

Alex

— Reply to this email directly, view it on GitHub https://github.com/amf71/ParallelGDDetect/issues/3#issuecomment-1589546803, or unsubscribe https://github.com/notifications/unsubscribe-auth/AZ2KM2CSNYYYVT3BYF3NH6TXLCBSNANCNFSM6AAAAAAY5NCYUE . You are receiving this because you authored the thread.Message ID: @.***>