FitDevo: accurate inference of single-cell developmental potential using sample-specific gene weight.

Area of Science:

• Computational Biology
• Genomics
• Developmental Biology

Background:

• Quantifying developmental potential is crucial for understanding cell differentiation and identifying key molecular markers in single-cell studies.
• Existing methods for inferring developmental potential from single-cell RNA sequencing (scRNA-seq) data have limitations.
• There is a need for robust and versatile computational tools to analyze developmental trajectories.

Purpose of the Study:

• To introduce FitDevo, a novel computational method for inferring developmental potential using scRNA-seq data.
• To demonstrate the effectiveness and superiority of FitDevo compared to current methods.
• To showcase the broad applicability of FitDevo across various biological contexts.

Main Methods:

• FitDevo generates sample-specific gene weights (SSGW) using a generalized linear model.
• It combines sample-specific information with gene weights trained on a large scRNA-seq dataset.
• Developmental potential is inferred by correlating SSGW with gene expression profiles.

Main Results:

• FitDevo was rigorously validated on an independent testing dataset comprising 28 scRNA-seq datasets.
• The method demonstrated superior performance compared to existing computational approaches.
• Successful applications were shown in deconvolution analysis of epidermis, spatial transcriptomics of heart and intestine, and breast cancer developmental potential analysis.

Conclusions:

• FitDevo provides a powerful and accurate approach for quantifying developmental potential from scRNA-seq data.
• The method's versatility makes it applicable to diverse research areas, including developmental biology, spatial transcriptomics, and cancer research.
• FitDevo offers a valuable tool for advancing single-cell data analysis and uncovering molecular insights.