ANC Workshop - Joshua Slaughter, Angus Chadwick
Tuesday, 3rd October 2023
Uncovering Gene-Deprivation Interaction with Targeted Learning - Joshua Slaughter
Abstract: Individuals experiencing higher levels of socioeconomic deprivation (SED) face a greater risk of adverse health outcomes, ultimately resulting in elevated mortality rates. This disparity is due to the complex interplay between genetic components and environmental factors, known as gene-environment (GxE) interaction. When examining complex phenotypes, it is crucial to understand how SED can influence biological mechanisms to uncover disease etiology, inform public health policy, and create targeted interventions. Fortunately, we are presented with a unique opportunity in the UK Biobank (UKBB), allowing us to jointly access genetic and socioeconomic variables. The UKBB provides two main summary measures for socioeconomic status in the Townsend Deprivation Index (TDI) and the Indices for Multiple Deprivation (IMD). Although the TDI and the IMD aim to provide metrics for SED and are used interchangeably throughout the literature, they have stark differences. In this project, we highlight the key differences between the TDI and the IMD at the level of definition, as well as how they categorize individuals within the UKBB cohort. We then aim to uncover the interaction effects between loci associated with body mass index (BMI) and type 2 diabetes (T2D) with these metrics for deprivation. We accomplish this using TarGene, a statistical workflow based in targeted learning that estimates targeted higher-order interaction effect sizes between genomic variants and environmental variables while avoiding parametric assumptions. Here, we perform several phenome-wide interaction studies (PheWIS) to characterize how SED interacts with genetic loci to impact health outcomes and uncover novel interaction effects between SED and genetic components on complex traits.
Linear dynamics for working memory - Angus Chadwick
Abstract: I will present some recent work on a novel method for optimisation of continuous-time linear dynamical systems, with application to a working memory task. In contrast to previous work which focused on attractors or feedforward dynamics, we find that high-dimensional rotational sequences are optimal for noise-robust working memory performance. This project is joint work with Laura Ritter (a previous UG project student).
Event type: Workshop
Date: Tuesday, 3rd October
Time: 14:00 (please note the unusual time)
Location: G.07
Speaker(s): Joshua Slaughter, Angus Chadwick
Chair/Host: Ajitha Rajan