ClassifyTE

ClassifyTE performs hierarchical classification of transposable elements using a stacking-based machine learning framework to assign TEs to taxonomic levels including the super-family for analyses of their genetic and evolutionary roles.

Key Features:

• Stacking-based machine learning: Leverages multiple machine learning techniques in a stacked configuration to improve classification robustness.
• Hierarchical classification: Assigns transposable elements to hierarchical taxonomic levels up to the super-family level.
• Benchmark training: Trains models on benchmark datasets to optimize classification performance.
• Performance optimization: Optimizes performance metrics such as the hF measure.
• Reported improvements: Demonstrated average percentage improvements of 4%, 10.68%, and 10.13% over several state-of-the-art methods according to reported hF comparisons.
• Homology-based validation: Validated against a new TE library generated by homology-based classification methods with high concordance at higher taxonomic levels.

Scientific Applications:

• TE taxonomy: Systematically categorizes transposable elements for taxonomic and comparative analyses at family and super-family levels.
• Genomic impact studies: Supports analyses of TE effects on gene expression, mutation rates, and genomic architecture.
• Evo-devo and evolution: Facilitates studies of germline and somatic evolution influenced by transposable element activity.
• Method benchmarking: Provides a reference for comparing TE classification accuracy across computational methods using hF and benchmark datasets.

Methodology:

Uses a stacking-based machine learning framework that leverages multiple ML techniques for hierarchical classification, trained on benchmark datasets, evaluated using the hF measure, and validated against a TE library generated by homology-based classification methods.