Hire Brett H. - AI Researcher

Developing and testing the utility of graph neural network (GNN) models to enable navigation and interpretability of complex networks within chemical and biological knowledge graphs, contributing to the team’s effort to improve prediction and interpretation of chemical-induced adverse effects.

* Worked in Computational Chemistry and Cheminformatics Branch (CCCB), performing research to assist with the Generalized Read Across (GenRA) project that aims to establish an algorithmic, reproducible approach to perform chemical read-across prediction * Worked towards incorporating information about chemicals metabolic transformations into GenRA approach * Applied graph theoretical approaches to chemical modeling in pursuit of quantitate measures of similarity between chemicals metabolic transformation networks * Employed NetworkX, pandas, numpy, pyTorch, pyTorch geometric, and other python packages to construct and train custom graph convolutional (GCN) deep learning architectures for analysis of graph representations of chemicals and metabolic transformation networks

* Employed graph analytical and database packages such as neo4j and NetworkX to perform research in various domains with the use of graph theoretical methods * Performed research on use of graph theoretical and statistical analysis methods on large datasets * Investigated a variety of graph embeddings methods that allow for the use of graph modeled objects in end-to-end machine learning pipelines * Maintained and developed the Graph Algorithms Pipeline for Pathway Analysis (GrAPPA) online platform that enables users to employ various graph theoretical and combinatorial tools to interpret high throughput biological data