Molecular Design TRIZ Cases
- Molecular Adsorption Simulation with Neural Network Potentials
This case uses neural network potentials and BFGS optimization to predict stable adsorption sites, angles, and energy distributions.
- AI Molecule Design Using Generative-Predictive Reinforcement Learning
Reinforcement learning guides neural networks to generate valid, diverse molecules with target properties.
- SAFE Molecular Strings for LLM-Compatible Scaffold Preservation
SAFE converts molecular structures into linked fragment strings that retain scaffold constraints for LLM-driven compound generation.
- SAFE Fragment Embeddings for Scaffold-Aware Molecular Language Models
SAFE converts molecular strings into order-agnostic linked fragments, preserving scaffolds for accurate language-model-driven design.
- Autonomous LLM Execution for Bio-Activity Discovery Pipelines
An autonomous language model navigates tech-bio tools to execute complex discovery workflows and retrieve accurate bio-activity data.