GeneMark-EP

GeneMark-EP integrates external protein evidence via the ProtHint pipeline to improve ab initio gene prediction and model parameter estimation in eukaryotic genomes.

Key Features:

• Integration with Protein Databases: Incorporates an external protein database as evidence by mapping proteins to the genome to inform gene prediction.
• ProtHint Pipeline: Uses ProtHint for large-scale protein-to-genome mapping and for extracting hints about splice sites and translation start and stop sites from protein alignments.
• Self-Training Mechanism: Employs iterative self-training that refines gene prediction model parameters using protein-derived hints.
• GeneMark-EP+ Coordinate Adjustment: In GeneMark-EP+ mode, adjusts predicted gene coordinates when they conflict with high-confidence protein-derived hints.
• Relation to GeneMark-ES and GeneMark-ET: Builds on the methodologies and self-training frameworks established by GeneMark-ES and GeneMark-ET.

Scientific Applications:

• Large eukaryotic genome annotation: Enhances gene prediction accuracy in large and complex eukaryotic genomes by incorporating protein evidence.
• Gene structure refinement: Improves identification of splice sites and translation initiation/termination sites to produce more accurate gene coordinates and structures.

Methodology:

Proteins are mapped to the genome with the ProtHint pipeline; mapping extracts hints on splice sites and translation initiation/termination; those hints are used to refine model parameter estimation via self-training; in GeneMark-EP+ predicted gene coordinates are adjusted when they conflict with reliable protein-derived hints.