JAX-ENT

Maximum Entropy Optimisation for Experimental Biophysics-Simulation Ensemble Integration using JAX.

Currently for HDX-MS data.

Supports:

To do: HDX-MS peptide data HDX-MS residue protection factor

Upcoming: SAXS NMR CryoEM???

##############################################################################

need to update the model parameters to be a pytree

• move generic infomration (temperature/ph up to the main settings class)

Optimiser: Each simulation should have its own experimental data section

• Should tie this together with the data splitting class

TODO: run_optimise -> seperate SGD into a sperate function -> extract optimiser step out -> map using jax

frame_average_features - need to find a way to make this jaxable - maybe is fine for now?

calc_BV_contacts_universe -> fix typing to use numpy - same with the rest of the featuriser code

create Enum to handle which parameters are being updated

##############################################################################

This package is aimed at bioinformatics researchers across a range of experience levels we aim to have a package that is batteries included but can also be extende to suit needs.

This package is centered around 3 main functions:

Featurise

Optimise

Analysis

CLI wrappers will be included under 'scripts'.

uses uv for package management

Please check that jax is correctly installed with CUDA/ROCm if desired.

To do: Add CUDA/ROCm flag to uv installation