Can I use SNP multiple sequence alignment matrix and a tree build based on it as inputs?

[CJJ8848]

Hi,

I created a SNP matrix only including variant sites presenting in all 72 samples and generated a tree based on the phy file converted from the SNP matrix using IQTREE. Then I input the fasta file converted from the phy file and the tree from IQTREE to CFML. It worked but removed so many recombination events for majority of branches. Could you please tell me whether I used the wrong input? I will be very appreciated!

[xavierdidelot]

No you can't do that. For ClonalFrameML to work your input fasta file needs to contain the non-polymorphic sites. This is needed for the detection of recombination. If you know the position of your SNPs along a reference genome you should be able to generate the full alignment as needed.

[CJJ8848]

Hi, Thank you so much for the quick response! I run CFML with my tree and the MSA containg both non-variant and variant sites in FASTA format last night and t works now! Could you please let me ask one more question? I am using perbranch model with a dispersion of 0.05. But when I try different dispersions, the tool gives very devergent recombination event results. Could you give me some suggestions about how to choose the right dispersion? The primary aim to use the perbranch instead of std model is that I would have a group of different recombination rate for each branch (for each sample included). Thank you very much!

jim

[xavierdidelot]

I would suggest to use the standard model, and you can get an estimate of the relative effect of recombination vs mutation r/m for each branch using the formula r/m=(t1-t2)/t2 where t1 is the branch length before correcting for recombination and t2 is the branch length after correcting for recombination.

[CJJ8848]

Hi,

Thank you very much for the detailed explanation! I would like to double check that the relative effect of recombination to mutation should equal to (ratio of recombination rate to mutation rate x recombination length x nu) as described in your paper. And here you mean I could calculate the ratio of recombination rate to mutation rate based on these two formulas? Like first calculate the r/m using t1 and t2 here and then calculate the branch-specific recombination rate using the formula given in the paper. Thank you very much!

Jim

Get Outlook for iOShttps://aka.ms/o0ukef

---

From: Xavier Didelot @.> Sent: Friday, June 2, 2023 10:54:24 AM To: xavierdidelot/ClonalFrameML @.> Cc: Cui, Jiajun @.>; Author @.> Subject: Re: [xavierdidelot/ClonalFrameML] Can I use SNP multiple sequence alignment matrix and a tree build based on it as inputs? (Issue #142)

⚠ Caution: External sender

I would suggest to use the standard model, and you can get an estimate of the relative effect of recombination vs mutation r/m for each branch using the formula r/m=(t1-t2)/t2 where t1 is the branch length before correcting for recombination and t2 is the branch length after correcting for recombination.

— Reply to this email directly, view it on GitHubhttps://github.com/xavierdidelot/ClonalFrameML/issues/142#issuecomment-1573465461, or unsubscribehttps://github.com/notifications/unsubscribe-auth/ALQIIZYKVA3QCUILLLAYRZLXJGZ5BANCNFSM6AAAAAAYXRW4QM. You are receiving this because you authored the thread.Message ID: @.***>

[xavierdidelot]

The two formulas estimate the same r/m. So you can use the one from my previous post instead of the one from the paper.