SiTaR

SiTaR predicts transcription factor binding sites (TFBSs) by scanning query sequences with input motifs to improve TFBS detection precision for promoter modeling and network inference.

Key Features:

• Alternative Approach to PWMs and HMMs: Each motif in the input set serves as a search template for scanning query sequences, avoiding preliminary motif alignment.
• Scoring System: Detected motifs are scored by assessing non-randomness through comparison of observed versus chance-expected motif counts and by accounting for the number of matching motifs and mismatches.
• Precision Enhancement: Demonstrates superior precision relative to PWM-based tools while maintaining comparable sensitivity and specificity, reducing false positive predictions.
• Base Composition Analysis: Calculates likelihoods of motif occurrences by considering the base compositions of both motifs and query sequences.

Scientific Applications:

• Promoter modeling: Improves accuracy of TFBS identification used in promoter architecture characterization.
• Network inference: Reduces false positives in TFBS calls to enhance reliability of inferred regulatory networks.

Methodology:

Scan query sequences using each input motif as a search template without preliminary alignment; score candidate sites by comparing observed versus chance-expected motif counts, by counting matching motifs and mismatches, and by estimating occurrence likelihoods from motif and query base compositions.