Optimal marker genes for c-separated cell types with SepSolve.

Area of Science:

• Genomics
• Computational Biology
• Bioinformatics

Background:

• Accurate cell type identification in single-cell RNA sequencing (scRNA-seq) relies on distinct marker gene expression profiles.
• Marker genes are crucial for targeted spatial transcriptomics, proteomics, and cell sorting.
• Traditional methods often test genes individually or oversimplify cell type variation.

Purpose of the Study:

• To develop a method for selecting a minimal set of marker genes that effectively distinguish all cell types simultaneously.
• To address limitations of existing methods, such as computational intractability and ignoring intra-cell-type variation.
• To identify genes that ensure cell types are 'c-separated' in selected expression dimensions.

Main Methods:

• Formulation of a linear program to jointly select marker genes.
• The method considers intra-cell-type expression variability without pairwise cell comparisons.
• Ensures 'c-separation' of cell types in the chosen gene expression space.

Main Results:

• Identification of a stable and small set of highly discriminative marker genes.
• The proposed linear programming approach is computationally efficient and scalable.
• The method accurately distinguishes between different cell types by accounting for expression heterogeneity.

Conclusions:

• The developed method provides an efficient and optimal solution for selecting marker genes in single-cell studies.
• This approach enhances the accuracy of cell type identification and facilitates downstream applications like spatial transcriptomics.
• The linear programming framework offers a robust way to handle expression variation within cell types.