README.md

In Cronjaeger/HotBirdHMM: Find Recombination Hoitspots using a Hidden Markov Model

HotBirdHMM

A project aiming to develop novel methods for finding recombination hotspots with a novel HMM-based approach.

Joint work w. Berklund, Terkelsen and Hobolth @BIRC (Aarhus Uni)

Overview of code:

• R/TransMat.R (written by Asger Hobolth) contains aux. functions for computing transition-rates of the hidden states in our HMM. This is done via matrix exponentiation using the 'expm' library.

• R/HMM.R Code for sampling from and computing likelihoods for our model of pairwise sequence heterozygosity.

Project Roadmap

1. Implement basics of HMM:

   1. Sampler [done]
   2. Likelihood computation (via forward algorithm) [done]

2. Work on maximum likelihood estimator

   1. Implement ML estimation of recombination rate [done]
   2. Test performance and power of method [initial testing done]
   3. Explore improvements to ML algorithm, if necessary (e.g. use a Baum–Welch/EM- type algorithm instead of Newtonian method currently employed) [currently not under consideration]

3. Search for hotspots

   1. Implement scanning algorithm [next step]
   2. Test on simulated data
   3. Apply to real data

Cronjaeger/HotBirdHMM documentation built on May 9, 2019, 6:01 p.m.