The portlet allows to submit a single instance of jModelTest to a remote Grid site.
The main aim of jModeltest is to select the best-fit model of nucleotide substitution for a given alignment of DNA sequences.
To accomplish these, jModelTest implements a computational pipeline that takes advantage of existing programs. In particular, most of the jModelTest run- time is taken by the likelihood calculations carried out with the program Phyml.
These calculations can take a variable amount of time depending on the data, number of candidate models and tree optimization. For each model in turn, jModelTest prepares a command line for Phyml, runs it and once it is finished it parses the Phyml output file.
Because jModelTest allows for the definition of up to 88 candidate models, jModelTest can repeat this whole process up to 88 times. Current version of the portlet submits a task containing a jModelTest instance (application plus input data) to the remote sites.
There, the application is executed on a serial way and the output data is recovered and presented to the user.