This library contains a minimal and vectorised implementation of the Li and Stephens model (haploid, forwards-backwards) as described in the 2022 kalis
paper.
- A simple formulation of the Li and Stephens model to showcase both the model itself and the PyTorch library.
- A general platform to investigate O(N^2) -> O(N) optimisations.
- Investigating automatic differentation of model parameters with respect to imputation/phasing accuracy.
- GPU support.
- Testing
see notebook.ipynb