Peptipedia v2.0

Peptipedia v2.0 facilitates peptide research and machine learning applications by serving as a comprehensive peptide sequence database with expanded functional activity classifications and predictive annotation capabilities.

Key Features:

• Expanded Sequence Collection: The database has grown by over 45% with peptide sequences that have reported biological activities.
• Enhanced Functional Activity Tree: The classification system incorporates new categories including cosmetic and dermatological activities, molecular binding, and anti-ageing properties.
• Binary Classification Models: More than 90 binary classification models trained and validated using protein language models and machine learning exhibit average sensitivities of 0.877 ± 0.0530 and specificities of 0.873 ± 0.054.
• Large-Scale Annotation: The classification models facilitate annotation of more than 3.6 million peptide sequences with unknown biological activities registered in the database.
• Description Tools: Tools based on structural and ontological properties enable peptide characterization.

Scientific Applications:

• Peptide Annotation: Machine learning models predict biological activities for peptide sequences with unknown functions.
• Biotechnological Research: The repository supports peptide study and annotation for therapeutic peptide discovery and development.
• Machine Learning Implementation: Integrated tools facilitate the application of machine learning strategies to study peptide sequences.

Methodology:

Peptipedia v2.0 utilizes protein language models and machine learning to train, validate, and incorporate binary classification models for peptide functional annotation.