MotifSampler

MotifSampler identifies over-represented sequence motifs in upstream regions of co-regulated genes using Gibbs sampling to infer position probability matrices for regulatory motif discovery.

Key Features:

• Gibbs Sampling Algorithm: Employs Gibbs sampling to probabilistically infer motifs by iteratively refining motif models and exploring sequence space.
• Position Probability Matrix Inference: Determines position probability matrices that represent discovered regulatory motifs.
• Higher-Order Background Models: Integrates higher-order nucleotide background models to account for dependencies among nucleotides beyond simple independence assumptions.
• Robustness to Noisy Data Sets: Optimized to improve motif detection in noisy datasets by combining Gibbs sampling with higher-order background models.
• Validation on Simulated and Real Data: Performance has been validated using simulated data and real biological datasets with well-characterized regulatory elements.
• Arabidopsis thaliana–specific Background Model: Uses a background model constructed from curated intergenic sequences for Arabidopsis thaliana to tailor motif detection to that species.

Scientific Applications:

• Transcriptome Analysis: Aids detection and clustering of co-expressed genes in transcriptome studies.
• Cis-regulatory Element Discovery: Enables discovery of shared cis-acting regulatory elements among co-regulated genes to inform gene regulation mechanisms.

Methodology:

Uses Gibbs sampling to iteratively infer position probability matrices for motifs and incorporates higher-order nucleotide background models to model sequence dependencies.