Author: Lars S Jermiin; Renee A Catullo; Barbara R Holland
Title: A new phylogenetic protocol: Dealing with model misspecification and confirmation bias in molecular phylogenetics Document date: 2018_8_27
ID: hr6wfx4g_14
Snippet: Having generated, say, = 1,000 pseudo-data, the next step involves finding the difference (δ) between the unconstrained (i.e., without assuming a tree and a model) and constrained (i.e., assuming a tree and a model) log-likelihoods (i.e., δ = ( ) − ( | , ), where is the data, is the tree, and is the model of sequence evolution). If the estimate of δ is greater for the real data than for the pseudo-data, then that result reveals a poor fit be.....
Document: Having generated, say, = 1,000 pseudo-data, the next step involves finding the difference (δ) between the unconstrained (i.e., without assuming a tree and a model) and constrained (i.e., assuming a tree and a model) log-likelihoods (i.e., δ = ( ) − ( | , ), where is the data, is the tree, and is the model of sequence evolution). If the estimate of δ is greater for the real data than for the pseudo-data, then that result reveals a poor fit between tree, model, and data (136). The approach described here works well for likelihoodbased phylogenetic analysis and a similar approach is available for Bayesian-based phylogenetic analysis (176). Parametric bootstrapping is computationally expensive and
Search related documents:
Co phrase search for related documents- difference find and δ difference find: 1
- δ estimate and δ likelihood: 1, 2, 3, 4, 5, 6, 7, 8, 9, 10
- δ estimate and log δ likelihood: 1, 2, 3
- δ likelihood and log δ likelihood: 1, 2, 3, 4, 5, 6, 7, 8, 9
- model tree and phylogenetic analysis: 1, 2, 3, 4, 5, 6, 7, 8, 9, 10
Co phrase search for related documents, hyperlinks ordered by date