ssolo
commented
2 years ago The numbers in question are estimates of the posterior mean number of events and as a result can be larger than unity.
E.g. if there is a 50% posterior probability of 0 Ds on a branch and 50% probability of 1 D one gets 0.5, but if there is 50% probability of 1 D and 50% of 2 Ds then one has 1.5.
I.e. the table at the end of the .uml_rec file summarises the number events per branch in the species tree from the reconciled gene trees above it (the long list of Newick strings). These reconciled gene trees are sampled according to their joint sequence and reconciliation likelihood (cf. eq. 3 here https://academic.oup.com/sysbio/article/62/6/901/1711882 ) in general they can have different topologies and reconciliations (i.e. series of DTL events).
In the example I attach (which I got by running ../build/bin/ALEml_undated Sab.tree Gab.tree.ale delta=0.2 tau=0.1 i.e. by forcing a higher probability of duplication for the purposes of this toy example ) in the list of reconciled gene trees you can see several alternative reconciled gene trees, e.g.
ones with 3 duplications on branch 3 of the species tree, these are the branches on the reconciled gene tree with @.***”:
@.**@*.**@*.***:1).4:0;
and also ones with only one duplication on branch 3 of the species tree @." again) and three transfers @.>3”, @.>3” and @.>c")
@.**@*.**@*.**@.>c:1).3:0;
etc.
the the estimate of the posterior mean number of D events on branch 3 of S is the average over these:
of Duplications Transfers Losses Originations copies singletons extinction_prob presence LL
.. S_internal_branch 3 1.75 0.43 0 0.47 3.05 0.3 9.31127e-07 0.93 -10.8095 ..
On 31 Oct 2022, at 03:51, 袁洋 @.***> wrote:
Dear Gergely Thanks for your explanations. I still can't understand what the 1 D and 2 Ds mean. I guess that 1 D and 2 Ds mean that a gene family can be duplicated once and twice on the branch of the species phylogeny, respectively. If it is true, how I should deal with these duplication events with inferred frequencies greater than 1? And If the inferred frequency is 1.9, I am not sure whether one or two duplication event should be counted.
Best regards, Yang Yuan
------------------ Original ------------------ From: "Gergely @.>; Date: Sun, Oct 30, 2022 05:31 PM To: @.>; Cc: @.***>; Subject: Re: Inferred frequencies greater than 1 (ALEml_undated in ALE 1.0 package)
Dear Yang Yuan,
The numbers in question are estimates of the posterior mean number of events and as a result can be larger than unity.
E.g. if ther is a 50% posterior probality of 0 Ds on a branch and 50% probability of 1 D one gets 0.5, but if there is 50% probality of 1 D and 50% of 2 Ds then one has 1.5.
Hope this answers the question.
Gergely
On 2022. Oct 29., at 16:55, 袁洋 @.***> wrote:
Hi, I am using ALEml_undated in ALE 1.0 package to infer the evolutionary history. No errors have been reported by the program. When I check the result files, I find that few inferred frequencies of duplications, transfers, losses, and originations are greater than 1. I am not sure if there's something wrong with it. I will use the threshold of 0.3 in the raw reconciliation frequencies to avoid misses of true events. Could I count these duplications, transfers, losses, and originations events with inferred frequencies greater than 1? I really need your help. If you could help, I really appreciate. I have appended the result files. Thanks very much.
Best regards, YangYuan
On 29 Oct 2022, at 17:00, 473021677 @.***> wrote:
Hi, I am using ALEml_undated in ALE 1.0 package to infer the evolutionary history. No errors have been reported by the program. When I check the result files, I find that few inferred frequencies of duplications, transfers, losses, and originations are greater than 1. I am not sure if there's something wrong with it. I will use the threshold of 0.3 in the raw reconciliation frequencies to avoid misses of true events. Could I count these duplications, transfers, losses, and originations events with inferred frequencies greater than 1? I really need your help. Thanks very much.
Best regards, YangYuan
— Reply to this email directly, view it on GitHub, or unsubscribe. You are receiving this because you are subscribed to this thread.
473021677
Hi, I am using ALEml_undated in ALE 1.0 package to infer the evolutionary history. No errors have been reported by the program. When I check the result files, I find that few inferred frequencies of duplications, transfers, losses, and originations are greater than 1. I am not sure if there's something wrong with it. I will use the threshold of 0.3 in the raw reconciliation frequencies to avoid misses of true events. Could I count these duplications, transfers, losses, and originations events with inferred frequencies greater than 1? I really need your help. Thanks very much.
Best regards, YangYuan