The ability to gather paired functional and spatial transcriptomic data would greatly improve our understanding of neuronal cell behavior and brain states. We aim to align data from in vivo functional imaging and post mortem fluoroscent imaging using a graph neural network approach. The goal of this project is to develop a tool to align 2D and 3D data for improved prediction of neuronal activity.

Our method, cSplotch, combines integrative and spatiotemporal generative modeling to improve the analysis of spatial gene expression data. It incorporates hierarchical experimental designs and spatial autocorrelation for more accurate inference of gene expression levels. The use of distinct tissue contexts allows for detection of reproducible changes in disease. This project will involve creating Bayesian models, optimizing computation, and adapting the pipeline for use with GPUs.

This project leverages spatial transcriptomics to study how colorectal tumors transition from canonical to non-canonical states during metastasis. It focuses on optimizing data visualization, tissue annotation, and machine learning algorithms to enhance the discovery of spatial tissue domains. Overall, the goal is to deepen our understanding of metastatic colorectal cancer and improve treatment strategies through advanced computational biology techniques.

This summer project focuses on enhancing microST, an open-source platform for spatial transcriptomics that uses cost-effective microfluidic devices to create modular DNA microarrays for high-throughput tissue analysis. It aims to re-engineer the fabrication workflow and develop new multiomics designs by integrating advanced 3D fabrication, microchip production, and rigorous quality control techniques. Ultimately, the project seeks to improve the accessibility and flexibility of spatial transcriptomics tools to address new biological challenges.

This project leverages brain-wide and age-related transcriptomic atlases to overcome traditional genetic tool limitations by enabling cell type- and state-specific manipulations through transcriptomic engineering. It utilizes RNA vector programming and both pooled and single-cell screens to identify sensors for aging-relevant cell types and states. Over the summer, the project will focus on developing essential molecular biology techniques and optimizing quality control workflows for a prime editing system.