Natural bitter blocking protein and peptide discovery using machine learning

Overview

Shiru Inc. is at the forefront of discovering natural bitter blocking proteins and peptides through a pioneering AI-driven approach. By leveraging proprietary protein databases and advanced machine learning models, Shiru identifies proteins that can bind to bitter receptors without activating them, effectively blocking bitter sensory components in foods such as alkalized cacao. This innovation aims to enhance flavor profiles and improve the sensory experience of food products.

Technical specifications

• Machine Learning Models: Utilizes Large Language Models (LLMs) to predict protein-protein interactions (PPI) and identify proteins that block bitterness.
• Validation Capabilities: Includes in vitro and biophysical assays to confirm bitter blocking activity.
• Receptor Testing: Comprehensive testing across 25 human bitter receptors to determine interaction efficiency.
• Screening Attributes: Evaluates physicochemical properties and regulatory risks such as solubility, thermal stability, allergenicity, and toxicity.

Technology readiness level

Currently at TRL 3, this technology is in the early stages of development, with ongoing validation processes in place to confirm efficacy and performance in laboratory settings.