STAMP

STAMP performs similarity searches and alignments of DNA-binding motifs to identify transcription factor (TF) associations and characterize motif relationships.

Key Features:

• Motif Querying and Alignment: Queries input motifs against extensive motif databases or user-provided datasets and aligns motifs using local/global and gapped/ungapped strategies with multiple comparison metrics.
• Multiple Alignments and Similarity Trees: Constructs multiple alignments, familial binding profiles, and similarity trees for collections of motifs to reveal evolutionary relationships and redundancies.
• Integration with Motif-Finders and Input Formats: Accepts output from 12 supported motif-finding tools and motif representations including frequency matrices, consensus sequences, and alignments of known binding sites.
• Optimization and Evaluation: Evaluates 105 combinations of comparison metrics and alignment algorithms and identifies local alignments combined with sum of squared distances or Pearson's correlation coefficient as effective for detecting eukaryotic DNA motif similarities.
• Generalized Binding Profiles and Clustering: Aligns and merges individual TF profiles to construct generalized binding models and employs an automatic method to determine the optimal number of clusters for TF classification.

Scientific Applications:

• Comparative transcriptional regulation: Enables comparison of DNA motifs to study transcriptional regulation across TFs and species.
• Microarray and ChIP-chip interpretation: Assists interpretation of microarray and ChIP-chip experiments by linking discovered motifs to known TF binding preferences.
• Target site prediction: Improves prediction of novel TF target sites by identifying motif similarity to characterized binding profiles.
• Evolutionary analysis and TF classification: Facilitates evolutionary studies of related motifs and refines TF family and subfamily classification through familial binding profiles.

Methodology:

Provides a flexible "mix-and-match" alignment platform using various comparison metrics, alignment methods (local/global, gapped/ungapped), multiple-alignment strategies and tree-building techniques; constructs multiple alignments, familial binding profiles and similarity trees; evaluates 105 metric/algorithm combinations (noting superiority of local alignments with sum of squared distances or Pearson's correlation for eukaryotic motifs); aligns and merges individual TF profiles and applies an automatic method to determine the optimal number of clusters; accepts frequency matrices, consensus sequences, alignments of known binding sites, and outputs from 12 motif-finders.