High Performance Machine Learning and Evolutionary Computing to Develop Personalized Therapeutics

To establish connectivity in the brain, neurons migrate from their birthplaces to final locations where they integrate into the neural circuitry. We study the facial branchiomotor neurons (FBMNs), which control facial expression and jaw movement in vertebrates. In zebrafish, the first FBMN to migrate acts as a pioneer – it actively explores the signaling environment of the brain and leaves a trailing axon behind it as a scaffold for followers. When the pioneer is absent, FBMN migration ceases. Using the single-plane illumination microscopy (SPIM), we are investigating the origins of FBMNs to understand how differences between pioneers and followers arise. We are adapting an automated pipeline for drift correction and cropping, plus cell segmentation and lineage reconstruction which will allow us generate cell tracks in a quick and reproducible manner. We are also pursuing novel data visualization methods which will allow us to edit FBMN lineages and migratory trajectories for accuracy.

Large-Scale Genome-Wide Enrichment Analysis of 31 Human Phenotypes

Genome-wide association studies (GWAS) aim to identify genetic factors that are associated with complex traits. However, individual genetic variants have small effects, making them hard to identify. In addition, lists of individual variant associations give limited biological insights. “Enrichment analyses” can address these problems by focusing on a set of genes, instead of individual genetic variants. Here we develop an efficient enrichment analysis method that jointly models GWAS summary statistics at millions of variants, and use it to analyse 3,913 biological pathways and 113 tissue-based gene sets in 31 human phenotypes. A key feature of our method is that inferred enrichment automatically informs new trait-associated genes. For example, enrichment in lipid transport genes suggests strong evidence for association between MTTP and low-density lipoprotein levels, whereas conventional analyses of the same data found no significant variant near this gene.

New Wine in Old Bottles - Ideological Transformation and Rhetorical Creation of Market in China’s People’s Daily, 1946 - 2003

One of the most surprising transformations in the 20th century was Communist China's embrace of market economy. To make it happen, there needs to be a huge ideological transformation in words to resolve the incompatibility between "free market" and orthodox Marxism. We attempt to model this transformation by applying computational content analysis to the full text of the communist party's mouthpiece, the People's Daily, from 1946 to 2003. There's a long debate about whether ideology has any independent explanatory power in historical processes. We found that the party's official rhetoric has progressed in a very linear and smooth fashion. Newness always comes out of old repertoires, and controversial concepts like "market economy" only became stabilized by attaching to an existing stable rhetorical subspace. Our findings echo the Weberian idea that ideology and culture should be viewed as a semi-autonomous social sphere that interacts with other social processes with its own logic.