K-mer-based machine learning method to classify LTR-retrotransposons in plant genomes

Area of Science:

• Genomics
• Bioinformatics
• Computational Biology

Background:

• Increasing availability of plant genome data necessitates advanced analytical tools.
• Current methods for detecting and classifying LTR retrotransposons are often manual, time-consuming, and lack individual accuracy.
• Existing bioinformatics tools have limitations in comprehensively analyzing LTR retrotransposons.

Purpose of the Study:

• To develop an automated and accurate method for classifying LTR retrotransposons in plant genomes.
• To improve the analysis of repetitive sequences within large-scale genomic datasets.
• To address the limitations of current semi-automatic and time-consuming LTR retrotransposon detection programs.

Main Methods:

• Utilized Machine Learning algorithms, specifically k-mer counts, for sequence classification.
• Developed a free-alignment approach for automated analysis.
• Trained and evaluated models on diverse genomic sequences.

Main Results:

• Achieved a high F1-Score of 95% for classifying LTR retrotransposons.
• Successfully differentiated LTR retrotransposons from other genomic sequences.
• Classified LTR retrotransposons into their respective lineages and families with high accuracy.

Conclusions:

• Machine learning, based on k-mer counts, provides an effective and automated solution for LTR retrotransposon classification.
• The developed method significantly enhances the efficiency and accuracy of plant genome analysis.
• This approach contributes to a more robust understanding of repetitive elements in plant genomics.