BiPad

BiPad predicts sequence motifs in unaligned sequences by modeling motifs as bipartite pairs of one-block position weight matrices with an associated gap distribution or as single contiguous position weight matrices to enable discovery of cis-regulatory and other genomic sequence elements.

Key Features:

• Bipartite Model Generation: BiPad generates bipartite models composed of a pair of one-block position weight matrices with an associated gap distribution.
• Single PWM Matrix Production: BiPad produces single contiguous position weight matrices (PWMs) for single-block motifs.
• Multiple Local Alignment (Entropy Minimization): BiPad performs multiple local alignments using an entropy minimization approach to identify high-information motif sites.
• Cyclic Refinement: BiPad applies cyclic refinement to iteratively improve alignment and motif accuracy.
• Stochastic Greedy Search Strategy: BiPad employs a stochastic greedy search strategy to explore sequence space for motif discovery.

Scientific Applications:

• Cis-regulatory element discovery: BiPad facilitates discovery and characterization of cis-regulatory elements and motif architectures in genomic sequences.
• Two-block motif analysis: BiPad enables identification and analysis of bipartite (two-block) motifs that involve separated binding sites with gap distributions.
• Nuclear receptor response element analysis: BiPad has been applied to identify human nuclear receptor response elements, including binding sites for HNF4alpha, CAR/RXR, and PXR/RXR.

Methodology:

Generation of bipartite models (pair of one-block PWMs with gap distribution) or single PWMs, multiple local alignment using entropy minimization, cyclic refinement, and a stochastic greedy search strategy.