Calum Pennington

Calum Pennington

Position

Enrichment Student

Cohort year

2024

Partner Institution

Bio

A computational engineer, Calum aims to add understanding and capability in biology by drawing on ideas from mathematics, computer science, and data science. Calum is doing a PhD in machine learning and computational genomics, which he began after a master's in computational biology (mathematical modelling and high-performance computing). His PhD is based at both a doctoral centre in quantitative methods for life sciences and the Dyson School of Design Engineering at Imperial College London.

Research interests

Calum is developing methods to represent complex biological data and uncover hidden structure in it, focusing on genomic data. One aim is to learn representations of data more automatically, and endow computers with an understanding of the system we're working on. 'Generative models' achieve this by learning to create new examples of data, which capture underlying patterns in the original data. Calum is developing a generative model that can generate realistic, synthetic genomic data. Another aim is to use geometric machine learning to perform learning on graph data. Deep neural networks have been optimised for data with Euclidean (grid-like) structure. But, organisms evolve by the acquisition of mutations, which can be captured by tree-like representations. So, we can represent genomic data also in non-Euclidean space, in the form of graphs - an approach that better captures relational patterns in biological data.