I just noticed that molecularevolution.org has provided, with Derrick Zwickl, a new online server for running phylogenetic maximum likelihood analyses using GARLI. The server for putting up new jobs (runs) can be found at:
A starting tree can be specified (although this doesn’t save time in GARLI like it does in PAUP*) and up to 2000 bootstrap reps can be performed.
This will prove a useful resource for anyone who doesn’t want to take time to put their jobs up on a supercomputer using some form of scripting (e.g. Perl), and for anyone who finds it disadvantageous to run phylogenetic algorithms on their personal desktop or laptop computers. I don’t know what this server is running or how fast the runs are, but GARLI (0.97) is very fast on the BYU supercomputer (“Mary Lou”), usually terminating in a matter of minutes when runs are placed on a single node. So, I’m expecting this service to be pretty quick; of course, the length of runs will depend on the number of taxa and characters, as well as usage. Note the option to generate site likelihoods, as in PAUP*, which will be useful for running analyses in CONSEL (although you may want to see the GARLI manual/web sites before generating these “site-likes” … click here for my previous post on GARLI with the appropriate links).