Version 2.4 is out as of January 5 2008: improvements in this version include calculations of marginal likelihoods for Bayes  Factors analyses (it is now possible the assess directly with MIGRATE whether two localities belong to the sample population or not); printing of effective sample sizes; presentation of the frequency of migration events over time and probability of which population contains the most recent common ancestor; memory foot-print for Bayesian analysis is considerably smaller; stability of parallel migrate has improved; For MacOS 10.5 users: I have created a new application migrateshell.app that sets up an environment that allows to run MIGRATE in parallel on multi-core machines without additional software installs.


Older News and Views: My 2006 paper about the comparison of Maximum likelihood and Bayesian inference got highlighted as a FAST-BREAKING PAPER in ISI/Thomson (here). Get your copy of the paper here.

MIGRATE-N

estimation of population sizes

and gene flow using the coalescent

Migrate estimates effective population sizes and past migration rates between n population assuming a migration matrix model with asymmetric migration rates and different subpopulation sizes. Migrate uses maximum likelihood or Bayesian inference to jointly estimate all parameters. The analysis can be constrained to subsets of migration patterns, such as setting some migration rates between populations to zero or constrain to symmetric rates or average over all migration rates, or use likelihood ratio tests and AIC values to compare different hypotheses. Migrate can use several different single or multilocus data types: sequence data using Felsenstein's 84 model with or without site rate variation, single nucleotide polymorphism data, microsatellite data using a stepwise mutation model or a brownian motion mutation model, and electrophoretic data using an 'k+1' allele model. The output comes in two flavors: PDF and TEXT file. The file can contain: Maximum likelihood: Estimates of all migration rates and all population sizes, assuming constant mutation rates among loci or a gamma distributed mutation rate among loci. Profile likelihood tables, Percentiles, Likelihood-ratio tests, and simple plots of the log-likelihood surfaces for all populations and all loci. (not all plots are implemented in the PDF, yet). Genealogy with best likelihood (Soon: printable through ET (eventtree -- http://popgen.scs.fsu.edu/Eventtree.html). Bayesian inference: Estimates of maximum posterior values of parameters and credibility intervals in table form, posterior distribution of parameters. Indications of convergence and effective sample size, presentation of frequency of migration events over time, genealogy with best likelihood printable through ET.Eventtree.htmlhttp://popgen.scs.fsu.edu/migabout.htmlshapeimage_2_link_0