HYPHY (pronounced as Hi-fi) is an abbreviation for [Hy]Pothesis Testing Using [Phy]logenies. It is a free multiplatform (Mac, Windows and UNIX) software package intended to perform maximum likelihood analyses of genetic sequence data and equipped with tools to test various statistical hypotheses. HYPHY was designed with maximum flexibility in mind and to that end it incorporates a simple high level programming language which enables the user to tailor the analyses precisely to his or her needs. These include relative rate and ratio tests, several methods of ML based phylogeny reconstruction, bootstrapping, model selection, positive selection, molecular clock tests, recombination, population structure tests and many more (for examples, see some of the published applications of HyPhy, via Google Scholar). HYPHY also comes with a collection of "standard" analyses which can be used as is.

The development of HyPhy started in 1997, with the first public release in 2000 and currently the package is converging to a 1.0 (non-beta) release with the most recent version being 0.99+ beta. We are planning to release a 1.0 version before the end of the year.

Analyze nucleotide, protein and codon sequences (in a variety of formats), including mixed data (e.g. intron exon data) and multiple data sets in one run.

Use predefined standard models, or define any reversible model you wish. Multiple models can be used in the same analysis, furthermore, tree branches of the same tree may have different models assigned to them.

HyPhy includes a versatile suite of methods to detect adaptive evolution at individual amino-acid sites and/or lineages, including generlizations of Nielsen-Yang (PAML) and Suzuki-Gojobori (ADAPTSITE) approaches and many others. See [datamonkey.org] and [GA Branch] for a free implementation running on our cluster.

Define arbitrary substitution models (e.g. for stem RNA, binary data) constraints on the parameters of the model(s).

Use HYPHY batch language to set up large and compilicted analyses and process the results.

Perform powerful and transparent parametric and non-parametric bootstrap simulations.

Choose from several standard preprogrammed analyses to get results very quickly or learn how to set up your own analyses by perusing and modifying example files provided with the distribution.

Use a feature-rich graphical user interface for Mac OS, Windows and X11 (GTK based) to set up analyses, define constraints, test hypothesis, visualize trees and results, define your own models, etc. (some examples).

The main difference between HYPHY and existing phylogeny mximum likelihood programs (PHYLIP, PAML, PAUP* and others) is its flexibility and customizability. The user is completely free to define any models, constrain parameters of the models, perform statistical tests and script complex and long (or repetitive) analyses via the batch language. HyPhy has a native graphical interface for all three platforms and is extensively parallelized, both for multiple processors (dual, quad machines) and cluster (MPI) environments.