GEM: Genome wide Event finding and Motif discovery
Website: http://cgs.csail.mit.edu/gem/
In the authors' own words:
GEM is a software tool to study protein-DNA interaction using ChIP-seq/ChIP-exo data. GEM links binding event discovery and motif discovery with positional priors in the context of a generative probabilistic model of ChIP data and genome sequence, resolves ChIP data into explanatory motifs and binding events at unsurpassed spatial resolution.
My summary:
A typical application of motif discovery is to identify short words (or patterns) of DNA sequence that indicate, e.g. where a DNA binding protein binds the genome. Typically the input to this class of programs is a set of short (e.g. 200bp) DNA sequences that come from a ChIP-seq or ChIP-chip experiment. The selection of the length and number of sequences if often very dependent on a whole series of decisions, such as threshold levels during the ChIP-seq "peak calling" step. Many web tools are often constrain the length and number of input sequences.
GEM is something new in motif discovery analysis, as it does not require decisions to be made about the sequences to analyse. GEM takes as input the mapped reads from a ChIP-seq experiment and set of control reads if these are present (which they should - my opinion). EDIT: It will then call binding events and discover motifs around the predicted binding sites.
Apart from identifying motifs of interest the tool will also identify patterns of spatial distribution between identified motifs. E.g. if three motifs are discovered GEM will identify patterns of binding between the primary motif (potentially homo-dimer binding) and compare the second and third motif to the primary motif, which can lead to the discovery of co-operative binding events.
I have just touched the surface, so look at the website and read the paper!
The authors are responsive and GEM is already on version 1.1. I at least hope this tool will continue to be developed.
