HMMgene

HMMgene predicts protein-coding genes in prokaryotic genomes using hidden Markov models to provide statistically scored identification of open reading frames (ORFs).

Key Features:

• Statistical Significance Estimation: Assigns statistical significance measures to putative genes to discriminate real genes from random ORFs.
• Hidden Markov Model Framework: Constructs HMMs tailored to each input genome by extracting a training set of genes using similarities to entries in Swiss-Prot.
• Automated Pipeline: Processes raw genomic data to generate a list of putative genes with associated significance scores.
• Performance and Accuracy: Reduces annotation errors such as incorrect start codons, missing genes, and over-annotation of spurious ORFs.
• Flexibility: Adjusts parameters based on the specific statistical properties of each genome to accommodate diverse prokaryotic genome types.

Scientific Applications:

• Comparative Genomics: Provides standardized gene predictions to enable consistent cross-genome comparisons.
• Large-scale Genome Annotation: Supports high-throughput annotation of newly sequenced prokaryotic genomes with statistical confidence scores.
• Microbial Genetics Research: Refines gene catalogs for prokaryotic genomes to improve downstream analyses in microbial genetics and functional inference.

Methodology:

Automatically extracts a training set of genes from the target genome using similarities to Swiss-Prot, estimates and applies a genome-specific hidden Markov model, and computes statistical significance scores for predicted ORFs to produce a ranked list of putative genes.