Tim Vivian-Griffiths

PhD Student - Cardiff University

I am currently in my final year of a PhD programme at the Institute of Psychological Medicine and Clinical Neuroscience at Cardiff University, funded by the Wellcome Trust. My project involves using machine learning algorithms, implemented in Python, to examine any possible patterns of association between genotyped genetic data and treatment-resistant diagnoses of schizophrenia.

The current trends in psychiatric genetics research is to use the results from Genome Wide Association Studies (GWAS) in schizophrenia to assign a single Polygenic Risk Score to an individual, and use this in a Logistic Regression model to attempt to predict case/control status. This is effectively amalgamating all of the genetic signals from across the genome down to a single score for each individual. Where my work differs is that I do not reduce the signal down in this way; instead, I use the GWAS results to create risk scores on the levels of either single nucleotide polymorphisms (SNPs - changes in single positions in the DNA), genes or sets of genes that have been shown in previous research to be functionally related. These can then act as different features to enter into machine learning models in an attempt to get a richer understanding of genetic processes, which could be related to the disorder.

Python specific interests:

On a daily basis, I make use of two popular libraries in Python, both of which build upon the popular numpy library:

Pandas - For all of the preparation steps from raw data to the arrays needed for the models.
scikit-learn - Used to build pipeline models for machine learning, making use of the models used for logistic regression and support vector machines.