Talk:Multispecies coalescent process

Untitled

Brief introduction to software for multispecies coalescent analyses:

86.22.89.241 changed this text:

"There are three basic methods for phylogenetic estimation in the multispecies coalescent framework: 1) full likelihood methods, which integrate over the uncertainty in gene trees (the approach originally proposed by Maddison 1997); 2) gene tree summary methods, which accept a set of estimated gene trees as input and output an estimate of the species tree; and 3) site pattern methods, which produce an estimate of the species tree directly from aligned sites."

To this text:

"There are two basic approaches for phylogenetic estimation in the multispecies coalescent framework: 1) full-likelihood or full-data methods which operate on multilocus sequence alignments directly, including both maximum likelihood and Bayesian methods, and 2) summary methods, which use a summary of the original sequence data, including the two-step methods that use estimated gene trees as summary input and SVDQuartet, which use site pattern counts pooled over loci as summary input."

I like the addition of "full-data" but don't think that combining summary tree methods like ASTRAL or MP-EST with SVDQuartets is a good idea. I suppose that calculating site pattern frequencies for multiple quartets of taxa can be viewed as summarizing the site pattern frequencies, but it is a very different type of summary from the two-stage species tree methods, where estimated gene trees are summarized. The methods are very different from an end user standpoint and lumping them together will confuse novice users. SVDquartets simply uses alignment data as input whereas ASTRAL, ASTRID, MP-EST, etc. require the user to estimate gene trees and then use those trees as input for the program that summarizes them. Perhaps 86.22.89.241 can explain the rationale for doing so. Regardless, some distinction should be added to avoid confusion in any end users that might come here for information. EBraun68 (talk)