GSoC_2015_project_hmms

Hidden Markov Models for gene imputation

Cleaning up Shogun's HMMS and implementing a gene imputation pipeline

Mentors

• Fabian Zimmer (github: qfma)
• Heiko (github: karlnapf, IRC: HeikoS)

Difficulty & Requirements

Easy/medium/advanced. You need to be able to

• get confused by C/C++
• trim beards
• count socks

Description

There are three major parts to this projects

• Cleaning up Shogun's HMM implementation (mostly to be done in pre-GSoC phase, see below)
• Extending CHMM with a simple MCMC sampler
• Build a workflow chain to impute genes

Details

In genetics unobserved genotypes are often imputed from a more complete dataset, see Imputation on Wikipedia. One of the most popular tools uses Markov chain Monte Carlo to accomplish this paper.

Fabian needs to put more details here.

Waypoints and initial work

• Step 1
• Step 2
• ...

Optional

Parts of the project that would be cool once the core is finished.

Why this is cool

Motivation to get involved here.

Useful ressources

• Imputation on Wikipedia
• MCMC for gene imputation paper
• Entrance tasks:
• Remove precompiler flags
• Unit test HMMs
• Document CHMM
• Clean up parallelisation
• Notebook on HMM