BioSeq-Diabolo: Biological sequence similarity analysis using Diabolo

Area of Science:

• Bioinformatics
• Computational Biology
• Genomics

Background:

• Biological sequence similarity analysis is crucial for predicting structure, function, and disease associations.
• Existing computational methods struggle with diverse data types and low sequence similarities (remote homology).
• Novel approaches are needed to accurately analyze biological sequence similarities.

Purpose of the Study:

• To apply natural language processing (NLP) semantics analysis techniques to biological sequence similarity analysis.
• To develop new concepts and methods for comprehensive and accurate biological sequence similarity analysis.
• To create a platform, BioSeq-Diabolo, for intelligent biological sequence similarity analysis.

Main Methods:

• Introduced 27 semantics analysis methods derived from NLP for biological sequence similarity analysis.
• Developed the BioSeq-Diabolo platform that accepts biological sequence data embeddings.
• Integrated biological sequence similarities using Learning to Rank (LTR) for supervised analysis.

Main Results:

• Demonstrated improved performance in protein remote homology detection, circRNA-disease association identification, and protein function annotation.
• Achieved better results compared to existing state-of-the-art predictors.
• The BioSeq-Diabolo platform intelligently identifies tasks and analyzes biological sequence similarities based on biological language semantics.

Conclusions:

• NLP-derived semantics analysis offers a powerful new approach to biological sequence similarity analysis.
• BioSeq-Diabolo provides an integrated platform for accurate and efficient biological sequence similarity analysis.
• The developed methods enhance predictions in key areas of biological research and diagnostics.