JIGSAW

JIGSAW integrates outputs from gene finders, splice site prediction programs, and sequence alignments to improve prediction of complete protein-coding genes in human DNA and other eukaryotic genomes.

Key Features:

• Integration of Multiple Evidence Types: JIGSAW combines diverse evidence sources—gene finders, splice site prediction programs, and sequence alignments—into its predictive models to improve accuracy over individual predictors.
• Adaptability for Eukaryotic Genomes: Implemented as part of a general-purpose eukaryotic gene-finding pipeline, JIGSAW has been adapted specifically for human DNA and can be applied to other eukaryotic genomes.
• Controlled Gene Finding Experiments: The methodology includes controlled experiments that assess the impact of different evidence types and feature states on predictive sensitivity and specificity.
• Evidence-Based Combiner Program: A dedicated "combiner" program leverages additional evidence tracks to achieve substantial gains in predictive accuracy and indicates that hidden Markov model enhancements may offer limited extra benefit.

Scientific Applications:

• Human gene prediction: Enhances prediction of complete protein-coding genes in human DNA by integrating multiple computational evidence types to increase sensitivity and specificity.
• Eukaryotic genome annotation: Supports annotation and analysis of other eukaryotic genomes and investigation of complex gene structures and functions through evidence-based gene models.

Methodology:

Integrates outputs from gene finders, splice site prediction programs, and sequence alignments; employs an evidence-based "combiner" program using additional evidence tracks; performs controlled experiments to evaluate the impact of evidence types and feature states; configured within a general-purpose eukaryotic gene-finding pipeline adapted to human DNA.