singIST: An integrative method for comparative single-cell transcriptomics between disease models and humans.

Area of Science:

• Computational biology
• Single-cell transcriptomics
• Drug discovery

Background:

• Disease models are crucial for drug discovery but imperfect approximations of human conditions.
• Existing computational methods for assessing model-human molecular resemblance lack single-cell resolution and explainability.

Purpose of the Study:

• To introduce singIST, a computational method for quantitative, explainable, and generalizable comparative single-cell transcriptomics analysis.
• To assess disease model similarity to human conditions at pathway, cell type, and gene levels.

Main Methods:

• Developed singIST, a unifying computational framework for single-cell transcriptomics analysis.
• Incorporated gene orthology, cell type presence, cell type/gene importance, and gene fold changes.
• Validated singIST using murine models of Atopic Dermatitis and human Hidradenitis Suppurativa data.

Main Results:

• singIST successfully recapitulated known biology in Atopic Dermatitis models and generated new hypotheses.
• Analysis of Hidradenitis Suppurativa revealed that CD3/CD28 stimulation selectively enhanced pathways already mirroring the human condition.
• Simulation studies confirmed singIST's robust performance and outperformed a baseline method.

Conclusions:

• singIST provides a powerful, explainable approach for evaluating disease models using single-cell transcriptomics.
• The method facilitates more accurate selection of preclinical models, improving drug discovery and development pipelines.