AI-Driven Protein Engineering and Molecular Simulations
X-Tour builds custom machine learning models and molecular simulation pipelines for protein stability analysis, aggregation prediction, and structure-based protein engineering.
Aggregation Prediction
ML classifiers to identify aggregation-prone regions and assess sequence-level developability
Molecular Dynamics
All-atom and coarse-grained MD simulations for conformational analysis and stability profiling
Custom ML Models
Domain-specific classifiers and predictive models trained on curated protein datasets
Structure Analysis
Leveraging AlphaFold and structure prediction tools for protein engineering insights
Prediction Meets Simulation
We combine machine learning with physics-based methods to solve problems in protein engineering — from sequence analysis to atomistic simulations.
ML for Protein Properties
We develop custom machine learning classifiers to predict protein behaviour directly from sequence and structure. Our models target properties that are critical for protein engineering and biotherapeutic development — trained on curated datasets and validated against experimental benchmarks.
- Aggregation-prone region (APR) identification
- Thermal stability and solubility prediction
- Developability assessment for biotherapeutics
- Protein language model fine-tuning on custom datasets
Molecular Dynamics & Force Fields
We run atomistic and coarse-grained molecular dynamics simulations to study protein behaviour at the molecular level. From conformational sampling to free energy calculations, our simulation pipelines provide mechanistic insight that complements and validates ML predictions.
- All-atom MD simulations for conformational analysis
- Enhanced sampling methods (replica exchange, metadynamics)
- Force field parameterization and validation
- Free energy calculations for binding and stability
Our Computational Pipeline
An end-to-end workflow from raw biological data to validated predictions — combining ML and physics-based simulation.
Data Curation
We aggregate and curate protein sequences, structures, and experimental measurements from public databases and proprietary datasets, with rigorous quality control at every step.
Model Development
Custom ML architectures — from gradient-boosted classifiers to protein language models — trained on domain-specific biological data with cross-validation and benchmarking.
Simulation & Validation
MD simulations and enhanced sampling methods provide atomistic-level validation. Predictions are systematically benchmarked against experimental data before deployment.
Deployment
Production-ready prediction tools and simulation workflows, deployable as APIs or integrated into existing research pipelines.
Where ML Meets Biophysics
We start from the biology, not the algorithm. Whether predicting aggregation hotspots or running microsecond-scale simulations, every decision is grounded in the underlying biophysics.
Research-Driven
Every model starts with deep understanding of protein biophysics. We study the thermodynamics, folding landscapes, and molecular mechanisms before designing the computational approach.
Physics-Informed
We combine the pattern recognition power of machine learning with physical principles — force fields, energy landscapes, and molecular dynamics — to build models that generalize beyond the training data.
Production-Grade
Our tools are designed for practical use from day one. Reproducible pipelines, well-documented APIs, and systematic benchmarking against experimental results.
Research & Perspectives
Exploring the intersection of AI, molecular simulation, and protein engineering.
Articles coming soon.