The inference of complicated cell identities can be improved using machine learning tools

 

By Carnegie Mellon University

January 13, 2023

                          Gene expression and activation result in cell patterns that are consistent in cell type and function throughout tissues and organs. Our knowledge of cells is enhanced by the discovery of these patterns, which has implications for identifying disease pathways.

The development of spatial transcriptomics technology has made it possible for researchers to study gene expression across whole tissue samples in their geographical context. But in order to make sense of this data and detect and comprehend these gene expression patterns, new computational techniques are required.

To close this gap, a research team lead by Jian Ma, the Ray and Stephanie Lane Professor of Computational Biology at Carnegie Mellon University, created a machine learning platform. The most recent issue of Nature Genetics included a cover article on their SPICEMIX technique paper.

Researchers can better understand the function that various spatial patterns play in a cell's total gene expression in complex tissues like the brain thanks to SPICEMIX. To achieve this, it uses spatial metagenes, which are collections of genes that may be associated with certain biological processes and might exhibit smooth or random patterns throughout tissue.

Ma, Benjamin Chidester, a project scientist in the department of computational biology, Tianming Zhou, and Shahul Alam, PhD candidates, worked with the team to utilise SPICEMIX to evaluate spatial transcriptomics data from mouse and human brain areas. To understand the variety of cell types and spatial patterns in the brain, they made use of SPICEMIX's special powers.

SPICEMIX distinguished spatial patterns of cell types in the brain more precisely than other approaches when applied to brain tissues. Through the mastered spatial metagenes, it also discovered novel brain cell types' expression patterns.

These results might provide a more comprehensive view of the intricacy of different types of brain cells, according to Zhou.

Spatial transcriptomics studies are increasing quickly, and SPICEMIX can assist researchers in making the most of this high volume, high dimensional data.

Ma added that his team's approach "has the potential to revolutionise spatial transcriptomics research and contribute to a deeper knowledge of both fundamental biology and disease development in complex tissues."