MetaProm: A Tool for Predicting Human Promoters
Author Information
Author(s): Wang Junwen, Ungar Lyle H, Tseng Hung, Hannenhalli Sridhar
Primary Institution: Center for Bioinformatics, University of Pennsylvania
Hypothesis
Can a neural network-based meta-predictor improve the prediction of alternative human promoters?
Conclusion
The MetaProm program outperforms individual promoter prediction programs in sensitivity and specificity.
Supporting Evidence
- The meta-predictor integrates predictions from existing programs to improve accuracy.
- 74% of 5' most promoters overlap a CpG island, indicating a strong association.
- The study evaluated 42,536 distinct human gene promoters on a genome-wide scale.
Takeaway
This study created a smart tool that helps find parts of our DNA that start making genes, which is important for understanding how genes work.
Methodology
The study evaluated six promoter prediction programs and developed a neural network-based meta-predictor using a large dataset of human gene promoters.
Limitations
The study's focus on alternative transcription start sites may underestimate the performance of the programs on traditional promoter predictions.
Digital Object Identifier (DOI)
Want to read the original?
Access the complete publication on the publisher's website