CellCover Defines Marker Gene Panels Capturing Developmental Progression in Neocortical Neural Stem Cell Identity.

Area of Science:

• Computational biology and bioinformatics
• Developmental neuroscience
• Genomics and transcriptomics

Background:

• Defining cell classes is crucial for analyzing single-cell RNA sequencing (scRNA-seq) data in biomedicine.
• Current marker gene identification relies on differential expression (DE) methods, which analyze genes individually.
• DE methods overlook gene redundancy and complementarity, limiting their ability to capture complex cell-class signals.

Purpose of the Study:

• To develop a novel method for identifying discriminating gene panels for cell classification.
• To overcome limitations of existing methods by analyzing multiple genes simultaneously.
• To explore conserved cell types and developmental trajectories across species using scRNA-seq data.

Main Methods:

• Proposed CellCover, a method framing gene panel selection as a minimal set-covering problem in combinatorial optimization.
• Applied CellCover to scRNA-seq data from developing mouse neocortex, mouse, primate, and human brains.
• Utilized transfer learning to assess marker identification across species and developmental stages.

Main Results:

• CellCover identified distinct cell-class-specific signals in the mouse neocortex compared to DE methods.
• Transfer learning demonstrated CellCover's ability to identify conserved neurogenesis markers and temporal progression markers.
• Transcriptomic signatures of human outer radial glia (oRG) appeared earlier in rodent precursors than in primates.

Conclusions:

• CellCover provides a powerful and efficient approach for identifying marker gene panels from scRNA-seq data.
• The method effectively captures cell-class-specific signals and conserved biological processes across species.
• Findings offer insights into the evolutionary origins of cell types critical for cortical expansion, like oRG cells.