St. Jude scMINER improves on single-cell data analysis

Single-cell analysis has unlocked gene expression data at unprecedented resolution, but making sense of that wealth of information is technically and computationally challenging. Current tools have low accuracy when identifying which genes “drive” transcription factors and signaling protein networks. A tool from the lab of Jiyang Yu, PhD, St. Jude Department of Computational Biology, called single-cell Mutual Information-based Network Engineering Ranger (scMINER), outperformed state-of-the-art algorithms in cell clustering, identifying transcription factor drivers and inferring signaling protein networks. The findings, highlighting the potential of scMINER to improve single-cell transcriptome analysis, were published in Nature Communications. 

“We have been developing scMINER for the past eight years. The key idea of scMINER is to use techniques in information theory to learn cell–cell or gene–gene relationships and translate them into powerful computational models that help expose ‘hidden’ signals from noisy single-cell omics data,” said corresponding author Yu.

“scMINER is an all-in-one solution for single-cell transcriptomics data analysis. It goes beyond the basic tasks like cell clustering and annotation — that most existing tools already offer —  and  takes it a step further by enabling cell type-specific network inference and hidden driver identification,” said leading co-first author Qingfei Pan, St. Jude Department of Computational Biology. “It also includes a user-friendly data visualization portal that makes result exploration both interactive and efficient.”

Read more