Binding Site Graphs for Predicting Transcription Factor Binding Sites
Author Information
Author(s): Timothy E Reddy, Charles DeLisi, Boris E Shakhnovich
Primary Institution: Boston University
Hypothesis
Densely connected subgraphs in a graph-theoretical framework can indicate transcription factor binding sites.
Conclusion
The proposed algorithm predicts yeast binding motifs significantly better than existing techniques and reduces false positive predictions to less than 30%.
Supporting Evidence
- The algorithm significantly outperformed 13 other TFBS prediction algorithms on yeast datasets.
- BSG predictions were robust to the choice and length of input promoters.
- The method reduced false positive predictions to less than 30%.
- BSGs predicted significant binding motifs for 118 TF–condition pairs.
Takeaway
This study introduces a new way to find where proteins attach to DNA, which helps scientists understand how genes are controlled.
Methodology
The study uses a graph-theoretical framework to analyze nucleotide co-occurrences and employs ensemble Gibbs sampling to construct binding site graphs.
Potential Biases
The reliance on empirical models may introduce biases that affect the generalizability of the findings.
Limitations
The algorithm's performance on non-yeast datasets was significantly worse, indicating a need for species-specific strategies.
Participant Demographics
The study focuses on yeast (Saccharomyces cerevisiae) as the model organism.
Statistical Information
P-Value
p<0.1
Statistical Significance
p<0.1
Digital Object Identifier (DOI)
Want to read the original?
Access the complete publication on the publisher's website