scFeatures: multi-view representations of single-cell and spatial data for disease outcome prediction

Area of Science:

• Computational Biology
• Genomics
• Bioinformatics

Background:

• Single-cell technologies generate complex, high-dimensional data.
• Current analytical methods lack consensus for summarizing sample-level information from single-cell data.
• There is a critical need for methods to extract meaningful insights from large-scale single-cell studies.

Purpose of the Study:

• To introduce scFeatures, a novel approach for creating interpretable sample-level representations of single-cell and spatial data.
• To demonstrate the utility of sample-level feature summarization for understanding disease mechanisms.
• To show the effectiveness of scFeatures in classifying disease status.

Main Methods:

• Development of the scFeatures R package.
• Application of scFeatures to summarize cellular and molecular features from single-cell and spatial omics datasets.
• Evaluation of summarized features for disease mechanism interpretation and classification.

Main Results:

• scFeatures generates interpretable cellular and molecular representations at the sample level.
• Summarized features are crucial for understanding disease mechanisms across diverse experimental studies.
• The approach accurately classifies disease status of individuals.

Conclusions:

• scFeatures provides a robust method for sample-level data summarization in single-cell and spatial omics.
• This approach enhances the interpretability and utility of single-cell data for biomedical research.
• scFeatures facilitates deeper insights into disease biology and patient stratification.