geneBasis

geneBasis selects optimal gene panels from single-cell RNA sequencing (scRNA-seq) data to enable targeted assays such as spatial transcriptomics and to identify markers for rare cell populations.

Key Features:

• Iterative gene selection methodology: Employs an iterative algorithm that adds genes to maximize the distance between the full scRNA-seq manifold and the manifold reconstructed from the currently selected gene panel.
• Optimization for designated panel size: Optimizes gene selection to achieve a user-specified number of genes for targeted panels.
• Multi-level evaluation: Evaluates selected gene panels at cell type, individual cell, and per-gene levels.
• Label-free selection and rare-cell marker identification: Performs selection without relying on pre-existing cell type labels and identifies markers for rare cell populations.
• Enhanced resolution of cell types and states: Improves resolution of distinct cell types and subtle cell-state differences relative to existing methods.

Scientific Applications:

• Developmental biology: Selects targeted gene panels from scRNA-seq to investigate cellular differentiation and developmental trajectories.
• Oncology: Facilitates identification of tumor cell types, microenvironment components, and state heterogeneity in cancer single-cell studies.
• Immunology: Enables panel selection to resolve immune cell types and activation states in immune profiling datasets.
• Neuroscience: Supports selection of genes to distinguish neuronal subtypes and states in brain single-cell studies.
• Spatial transcriptomics integration: Produces gene panels compatible with spatial transcriptomics and other targeted assays to map cell types and states in tissue context.

Methodology:

Iterative gene selection that maximizes the distance between the full scRNA-seq data manifold and the manifold built from the selected panel; optimization constrained to a user-specified gene count; evaluation performed at cell type, individual cell, and per-gene levels; selection operates without pre-existing cell type labels to enable rare-cell marker identification.