PhyloGibbs-MP (version 2.0)

PhyloGibbs is a motif finder to find binding sites for transcription factors in cis-regulatory sequences of DNA.

PhyloGibbs-MP is a new version, which is now feature-complete and mostly bug-free. (This is a late beta version; the final version will be released Real Soon Now.)

REFERENCE: Rahul Siddharthan, "PhyloGibbs-MP: Module Prediction and Discriminative Motif-Finding by Gibbs Sampling", PLoS Comput Biol 4(8): e1000156 (2008) (link) (supporting data)

PhyloGibbs-1.0 is now deprecated, and I will upload binaries for PhyloGibbs-MP soon (user-contributed binaries are welcome.) The new program does almost everything PhyloGibbs-1.0 did (a few obscure options have been removed). New features include:

• A significant speed improvement (it runs about 10x faster, without sacrificing much accuracy)
• Instead of picking a fixed prior expected number of binding sites, one now chooses a prior density: this density applies to each sequence and is independent of the relatedness of sequences. See the -p and -n options.
• One can restrict predictions to an unspecified subregion of the given input sequence, for example when one is unsure of the location of a cis-regulatory module. In other words, PhyloGibbs-MP can predict CRMs simultaneously with binding sites. See the -l and -n options.
• One can predict motifs that discriminate between two or more groups of differently regulated genes (that is, motifs that appear preferentially in one group). See the -d option.
• The implementation of the -M option (informative priors) has been changed.
• Optionally, PhyloGibbs-MP outputs annotations that can be read by the Generic Genome Browser. See the -g option.
• Many other details have changed: please read the manpages carefully!

PhyloGibbs-MP 2.0 is now available. It contains only minor bugfixes from the previous version (of June 30, 2008).

The code

For historical reasons, source code of June 30, 2008 (used for benchmarks in the paper) is here. (No binaries yet.)