Joshua Welch, Ph.D., an associate professor in the Department of Computational Medicine and Bioinformatics (DCMB) was awarded a Chan Zuckerberg Initiative (CZI) grant for his project titled: “Atlas-scale Hierarchical Identification of Cell Types and Functions.”

With this new funding, the Welch lab will develop an AI approach for identifying the type of a cell from its gene expression. While many previous approaches for this problem exist, theirs is unique in its ability to use both general and specific cell type annotations to calibrate its predictions. Another unique aspect of this project is in predicting the functions that cells can perform based on gene expression. The ability to predict these properties would save much time and standardize the vocabulary scientists use when analyzing single cell RNA sequencing (scRNA) data.

“We are excited to work toward better cell type classification algorithms, and we're grateful for both the funding and the infrastructure that CZI has developed that makes this project possible.”—Joshua Welch, Ph.D.

This project depends on the availability of many high-quality scRNA datasets with manually curated labels. The Welch team plans to use CZI's CZ CELLxGENE Discover Census dataset, which is the largest single collection of scRNA data in which every cell is labeled with a cell ontology term. 

This project will use the latest neural network approaches to develop an AI approach that can label cells. The key challenge for this project is to address the laborious, subjective, and error-prone nature of manually labeling cell types. The latest AI models provide a promising solution to this challenge.

Welch and his Ph.D. candidate Jialin Liu already received a grant from CZI in 2023 for developing two novel computational tools that detect signaling among single cells from spatial transcriptomic data. Read more.

Congratulations, Dr. Welch!