GlimmerM, Exonomy and Unveil: three ab initio eukaryotic genefinders

Area of Science:

• Bioinformatics
• Computational Biology
• Genomics

Background:

• Accurate gene prediction is crucial for understanding eukaryotic genome function.
• Existing gene finders have limitations, necessitating the development of novel approaches.

Purpose of the Study:

• To introduce and evaluate three novel ab initio gene prediction programs: Exonomy, Unveil, and GlimmerM.
• To assess the performance of these programs individually and as an ensemble.

Main Methods:

• Exonomy: Generalized Hidden Markov Model (GHMM) with 23 states.
• Unveil: Standard Hidden Markov Model (HMM) with 283 states.
• GlimmerM: Decision trees and Interpolated Markov Models (IMMs).
• All programs are re-trainable for different organisms.

Main Results:

• Performance evaluation on Arabidopsis thaliana.
• Each gene finder demonstrated cases of outperforming the others.
• The ensemble approach showed collective value, suggesting improved overall prediction accuracy.

Conclusions:

• The developed gene prediction programs (Exonomy, Unveil, GlimmerM) offer valuable tools for eukaryotic genome analysis.
• Utilizing an ensemble of these gene finders can enhance the accuracy and robustness of ab initio gene prediction.
• These programs are accessible via web servers, facilitating their use in the broader research community.