HMRPred: A Machine Learning-Based Web Resource for Identification of Heavy Metal Resistance Proteins

Area of Science:

• Environmental Microbiology
• Biotechnology
• Bioinformatics

Background:

• Heavy metal contamination threatens ecosystems and agriculture.
• Heavy metal resistance (HMR) proteins are crucial for organism survival against toxic metals.
• Existing methods struggle to identify diverse HMR proteins across different organisms.

Purpose of the Study:

• To develop a machine learning framework for identifying HMR proteins across ten key metals.
• To create a comprehensive resource for classifying HMR proteins, addressing limitations of current methods.
• To support research in environmental remediation and synthetic biology.

Main Methods:

• Developed HMRPred, a machine learning framework utilizing curated protein datasets.
• Extracted sequence-derived features including amino acid composition and physicochemical descriptors.
• Trained and optimized classifiers for each metal using various machine learning algorithms.

Main Results:

• Achieved high performance with AUC-ROC >98% in cross-validation and independent testing for HMR protein identification.
• HMRPred successfully classifies HMR proteins across ten metals: arsenic, cadmium, chromium, copper, iron, lead, mercury, nickel, silver, and zinc.
• The framework is accessible as a web resource for researchers.

Conclusions:

• HMRPred provides a reliable method for identifying heavy metal resistance proteins.
• Facilitates genome-wide annotation of metal resistance determinants.
• Supports advancements in bioremediation, environmental microbiology, phytoremediation, and synthetic biology.