TOUCAN: a framework for fungal biosynthetic gene cluster discovery

Area of Science:

• Biotechnology
• Bioinformatics
• Genomics

Background:

• Fungal secondary metabolites (SMs) are crucial for pharmaceuticals, but discovering their biosynthetic gene clusters (BGCs) is challenging due to genomic diversity.
• Existing BGC discovery methods have limitations, hindering the identification of novel fungal SMs.

Purpose of the Study:

• To introduce TOUCAN, a supervised learning framework for enhanced fungal BGC discovery.
• To enable BGC prediction directly from amino acid sequences for broader applicability.

Main Methods:

• Developed a supervised learning framework (TOUCAN) utilizing expert-curated datasets.
• Integrated functional, evolutionary, and compositional features with advanced classifiers.
• Implemented robust post-processing techniques for accurate BGC identification.

Main Results:

• TOUCAN demonstrated superior performance compared to previous BGC discovery approaches.
• Achieved a high F-measure of 0.982 on BGC regions within the *Aspergillus niger* genome.
• The framework's ability to process amino acid sequences facilitates analysis of uncurated genomic data.

Conclusions:

• TOUCAN offers a significant advancement in fungal BGC discovery, improving efficiency and scope.
• The framework's adaptability allows for expansion to other species and feature sets.
• Facilitates the discovery of novel bioactive compounds for potential pharmaceutical benefits.