Identifying gene expression programs of cell-type identity and cellular activity with single-cell RNA-Seq

Area of Science:

• Computational Biology
• Genomics
• Neuroscience

Background:

• Understanding cell-type identity and activity gene expression programs is vital for tissue organization.
• Single-cell RNA sequencing (scRNA-Seq) data presents challenges in disentangling these mixed expression profiles.

Purpose of the Study:

• To benchmark and enhance matrix factorization methods for inferring distinct gene expression programs.
• To develop a computational approach for separating cell identity and activity programs in scRNA-Seq data.

Main Methods:

• Benchmarking matrix factorization techniques, including the development of consensus non-negative matrix factorization (cNMF).
• Simulations were used to validate the accuracy of cNMF in inferring programs and their cellular contributions.
• Application of cNMF to scRNA-Seq datasets from brain organoids and visual cortex.

Main Results:

• cNMF accurately infers both cell identity and activity programs, along with their relative contributions within individual cells.
• The method refines existing cell type classifications and identifies expected activity programs like cell cycle and hypoxia.
• Novel activity programs potentially related to neurosecretory phenotypes and synaptogenesis were discovered.

Conclusions:

• cNMF provides a robust computational framework for dissecting complex gene expression patterns in scRNA-Seq data.
• This approach enhances the resolution of cell-type identification and uncovers new insights into cellular functions and processes.
• The method has broad applicability for analyzing diverse scRNA-Seq datasets in biological research.