Software & methods
What the lab has released publicly, and the tools it works with day to day. Everything in the first list is open source and installable without asking us for anything.
Public releases
Software
Most of the lab’s code is not public yet. Rather than list private repositories that nobody can open, this page shows only what is genuinely available, and grows as more is released.
ml-guide-hdac
MIT licence
A practical guide to building machine-learning models for histone deacetylase inhibitor prediction, written for people in the lab and left public because it is useful outside it too.
What we use
Methods and tools
The lab is deliberately not tied to one framework. What follows is what actually appears in the lab’s work, grouped by the question it answers.
Predicting
Molecular property and activity models in PyTorch, scikit-learn, XGBoost and LightGBM; molecular representations from RDKit and the Uni-Mol pretrained models; SHAP to interpret what a trained model is actually keying on.
Docking and screening
AutoDock Vina for pose prediction and virtual screening, with Open Babel for format and protonation handling, and AlphaFold where an experimental structure is not available.
Simulating
Molecular dynamics in GROMACS and AMBER, with ACPYPE for ligand parameterisation and OpenMM where a Python-native engine is easier to script.
Quantum chemistry
ORCA, Gaussian 16 and xTB for the cases where a force field is not trustworthy — charges, conformer energies and reaction-relevant geometries.
Making and measuring
Synthesis of hydroxamic acid and oxoindoline-based scaffolds, followed by enzyme inhibition and cytotoxicity assays. The published record is the evidence here: twenty of the lab’s papers report compounds synthesised, and twenty-three report biological evaluation.
Looking at structures
PyMOL for inspection and figures, because a trajectory that has not been looked at is a number nobody should trust yet.