EasyGene

EasyGene predicts protein-coding genes in prokaryotic genomes using a hidden Markov model (HMM) and Swiss-Prot–derived training sets to assign statistical significance to open reading frames (ORFs) and standardize annotations.

Key Features:

• Automated Gene Prediction: Identifies genes by analyzing open reading frames (ORFs) and generating a list of putative genes with associated likelihoods.
• Statistical Significance Assessment: Uses a hidden Markov model (HMM) to evaluate the probability of an ORF being a genuine gene, considering ORF score and length.
• High-Quality Training Set Utilization: Automatically extracts a training set of genes from the genome using similarities in Swiss-Prot to estimate HMM parameters.
• Performance Evaluation and Comparison: Has been tested on 143 prokaryotic genomes and identified discrepancies between existing annotations (GenBank, RefSeq) and its predictions, including over-annotation in GC-rich genomes and under-annotation.
• Standardized Annotation Method: Provides a consistent annotation approach that facilitates transparent genome comparisons and has shown slight performance gains over existing annotations in evaluations.
• Automated Pipeline: Implements an automated computational pipeline from raw genome input to generation of statistically scored lists of putative genes.

Scientific Applications:

• Genome Annotation: Produces standardized gene annotations for prokaryotic genomes.
• Comparative Genomics: Enables consistent cross-species or strain comparisons by applying uniform annotation criteria.
• Error Correction: Detects annotation errors such as incorrect start codons, over-annotated ORFs, and missing genes.

Methodology:

Parameters are estimated by training a hidden Markov model (HMM) on genome-derived gene sets extracted via Swiss-Prot similarity; ORFs are scored by length and HMM-derived probabilities to produce lists of putative genes with statistical significance.