tag:blogger.com,1999:blog-8499895524521663926.post1407645148148445094..comments2024-11-06T04:50:42.713-05:00Comments on Phylogenetic Tools for Comparative Biology: Addendum to estimating ancestral states under an OU modelLiam Revellhttp://www.blogger.com/profile/04314686830842384151noreply@blogger.comBlogger1125tag:blogger.com,1999:blog-8499895524521663926.post-50318508050514927622013-05-24T08:28:35.256-04:002013-05-24T08:28:35.256-04:00Hi Liam,
thanks for this post! My comment is prob...Hi Liam,<br /><br />thanks for this post! My comment is probably a bit late, but I wonder: how (and why!) does ouTree work at all? <br /><br />I understand that it is transforming branch lengths so that the ML estimate for standard BM will return ancestors under the OU model - but how is that possible?<br /><br />I can imaging that for a character that grows, say, quadratically or exponentially over time, you modify branches by lengthening them appropriately, and I can somehow imagine it for Pagels lambda etc., but I don't get it for OU.<br /><br />I looked into the code and it does something like scale each branch exponentially according to its length and height in the tree and according to alpha. But I wonder: there is no theta involved in that formula. <br /><br />I know that the tip values under OU are distributed according to a normal distribution, so in that sense it is similar to a BM model. But while the mean of that Normal under a BM is simply the value of the root, the mean of the Normal under an OU model is dependent on theta. <br /><br />So I wonder - how can scaling the branches and using ML inference be enough to infer ancestral states under OU - since (I believe) this scaling can only influence the variation, but not the mean? Is there a manuscript somewhere that explains the math behind this?<br /><br />I'm sure I'm missing something very basic here, but I can't figure it out. <br />It seems that transforming the tree and then using standard ML under BM is a standard way to find a ML estimate under more complex models - but how does this magic work?<br /><br />Thanks a lot!<br />Best,<br />Christina<br /><br /><br />Anonymousnoreply@blogger.com