SGP2

SGP2 integrates sequence similarity searches and ab initio gene prediction to improve identification of conserved mammalian genes through comparative analysis of human and mouse genomic sequences.

Key Features:

• Integration of Methods: Combines ab initio gene prediction programs with TBLASTX protein-level sequence similarity searches between genomes.
• Cross-Species Comparison: Performs pairwise human–mouse comparative analysis exploiting protein-level conservation to refine gene models.
• Benchmark Performance: Demonstrates increased specificity and sensitivity relative to traditional ab initio methods on single-gene datasets, curated chromosome 22 predictions, and ENSEMBL whole-genome predictions.
• Versatility with Data Types: Operates on fully assembled genomes and fragmented shotgun sequencing data (3x coverage) without compromising prediction quality.
• Experimental Verification: Predictions have been validated by RT-PCR amplification and direct sequencing, with a 76% verification rate for non-overlapping mouse gene predictions.

Scientific Applications:

• Novel Gene Discovery: Identification of over a thousand gene predictions that do not overlap known genes in mammalian genomes.
• Homolog Identification: Detection of new homologs including homeobox proteins and aquaporins.
• Functional and Developmental Studies: Supports discovery of genes with restricted expression patterns and genes lacking homologs in fish genomes relevant to developmental biology.

Methodology:

Two-stage procedure: initial ab initio gene model prediction followed by TBLASTX searches between human and mouse genomes to identify conserved protein sequences and refine predictions.