The OGCleaner: filtering false-positive homology clusters

Area of Science:

• Bioinformatics
• Computational Biology
• Genomics

Background:

• Detecting homologous sequences is crucial for predicting protein structure/function, gene annotation, and phylogenetic tree construction.
• Current heuristic methods for homology cluster quality control often focus on pairwise comparisons, neglecting cluster-wide analysis.
• This limitation can impact the accuracy of downstream analyses, particularly with noisy data.

Purpose of the Study:

• To introduce the Orthology Group Cleaner (OGCleaner), a novel tool for filtering putative orthology groups.
• To enhance the accuracy of homology and non-homology cluster identification by considering all sequences within a cluster.
• To improve the quality of phylogenetic tree construction, especially when dealing with lower-quality transcriptome assemblies.

Main Methods:

• The OGCleaner utilizes machine learning algorithms trained on high-quality orthologous groups from OrthoDB.
• It distinguishes between true-positive and false-positive homology groups by examining the entire cluster.
• The tool is implemented as a package for broader accessibility and application.

Main Results:

• The OGCleaner effectively filters putative orthology groups, improving cluster quality.
• It provides a more comprehensive approach to homology detection than pairwise methods.
• The tool demonstrates potential for enhancing the reliability of phylogenetic analyses.

Conclusions:

• The OGCleaner offers a robust solution for refining orthology group identification.
• By analyzing clusters holistically, it addresses limitations of existing heuristic methods.
• This tool is expected to significantly benefit genomic and phylogenetic research, particularly in challenging datasets.