PromFD

PromFD predicts vertebrate RNA polymerase II promoters from primary DNA sequences to support promoter annotation and reduce false-positive promoter calls.

Key Features:

• Algorithmic Innovation: Leverages patterns and Information Matrix Database (IMD) matrices over-represented in known vertebrate promoters to improve prediction accuracy and lower false positives.
• Data Sources and Training: Uses promoter sequences from the Eukaryotic Promoter Database and non-promoter sequences from GenBank, divided into training and test sets.
• Pattern Recognition: Systematically searches for short string patterns (5-10 base pairs) and IMD matrices that are over-represented in promoters and stores these findings in the PromFD database.
• Scoring and Prediction: Scores input DNA sequences based on their content of identified database entries to predict promoter locations and potential TATA box sites.
• Performance Metrics: Reported detection rates are 71% in the training set with a false-positive rate below 1 per 13,000 bp and 47% in the test set with a false-positive rate below 1 per 9,800 bp.
• Programming Language: Source code is written in 'c+2'.

Scientific Applications:

• Genome Annotation: Supports annotation of newly sequenced vertebrate genomes by identifying RNA polymerase II promoter regions.
• Large-scale Genomic Studies: Enables large-scale promoter prediction with reduced false positives for analyses across extensive DNA sequences.
• Gene Regulation Analysis: Aids investigation of gene expression and regulatory element identification by locating promoter regions and potential TATA boxes.

Methodology:

PromFD uses Eukaryotic Promoter Database and GenBank sequences split into training and test sets, systematically searches for 5–10 bp string patterns and IMD matrices over-represented in promoters, stores matches in the PromFD database, and scores input DNA sequences by their content to predict promoter locations and potential TATA box sites.