Discriminative motif discovery in DNA and protein sequences using the DEME algorithm

Area of Science:

• Bioinformatics
• Computational Biology
• Genomics

Background:

• Motif discovery identifies conserved patterns in biological sequences.
• Discriminative motif finders enhance sensitivity by differentiating sequence sets.
• Applications include transcription factor binding sites and protein thermal stability motifs.

Purpose of the Study:

• To introduce DEME, a novel discriminative motif discovery algorithm.
• To evaluate DEME's performance on synthetic and biological datasets.
• To assess DEME's ability to find biologically relevant motifs.

Main Methods:

• DEME utilizes a probabilistic model for motif representation.
• It employs a hybrid global and local search strategy.
• Incorporates a Bayesian prior for protein motif columns, leveraging residue characteristics.

Main Results:

• DEME outperforms non-discriminative methods with decoy or variant motifs in negative sequences.
• It performs comparably to non-discriminative algorithms for yeast transcription factor binding motifs.
• DEME successfully identifies informative protein motifs related to thermal stability.

Conclusions:

• DEME is a valuable tool for discriminative motif discovery, particularly in challenging scenarios.
• The algorithm demonstrates utility in both DNA and protein sequence analysis.
• DEME is available as a free, stand-alone program for academic use.