This repository contains the code for Chapter 3 of my PhD thesis, entitled Histone Modification Occupancy Prediction Through Language Modelling.
The Hidden Markov Model code is based on this implementation of HMMs in Pytorch, but with several bug fixes and various other improvements.
The supervised PCFGs were evaluated using Mark Johnson's CKY algorithm implementation. The unsupervised PCFGs were built using Kim et al's neural and compound PCFG code.