FirstEF

FirstEF predicts promoters and first exons by applying discriminant functions that recognize structural and compositional features such as CpG islands, promoter regions, and first splice-donor sites.

Key Features:

• Discriminant functions: Uses a set of discriminant functions to detect patterns associated with promoters and first exons.
• Decision tree framework: Integrates the discriminant functions into a decision tree to enhance predictive classification.
• CpG and non-CpG exon identification: Specifically distinguishes CpG-related first exons from non-CpG-related first exons.
• Feature recognition: Detects structural and compositional features including CpG islands, promoter regions, and first splice-donor sites.
• Performance metrics: Cross-validation shows prediction of 86% of first exons with a 17% false positive rate.

Scientific Applications:

• Genome-level evaluation: Applied for large-scale genomic analyses and genome-wide exon prediction assessments.
• Chromosomes 21 and 22 analysis: Tested on the finished sequences of human chromosomes 21 and 22.
• Experimental comparison: Predictions were compared with experimentally verified first exon locations.
• Whole-genome analysis: Employed in an analysis across all 24 human chromosomes.

Methodology:

Integrates discriminant functions into a decision tree framework to identify CpG-related and non-CpG-related first exons, detects CpG islands, promoter regions, and first splice-donor sites, and uses cross-validation to estimate predictive performance.