A Composite interaction Score: Prioritizing cell-cell interactions from single-cell RNA-seq with application to...

Area of Science:

• Computational biology
• Genomics
• Immunology

Background:

• Cell-cell interactions (CCIs) are crucial for tissue health and immune responses.
• Current computational tools for inferring CCIs from single-cell RNA sequencing (scRNA-seq) data lack a standardized framework for result prioritization.
• Existing consensus strategies often fail to identify the most biologically significant CCIs.

Purpose of the Study:

• To develop a novel ranking strategy for CCIs that emphasizes reproducibility and biological relevance.
• To apply this strategy to investigate CCIs within epithelial barrier tissues.

Main Methods:

• Introduction of the Composite Interaction Score (CIS), a consensus metric integrating predictions from six CCI inference tools using the LIgand-receptor ANAlysis (LIANA) package.
• Implementation of ranked-biased precision to weight tool agreement, prioritizing top-ranked concordant interactions.
• Benchmarking CIS against a naive average-rank method using perturbed datasets to assess precision and recall.

Main Results:

• CIS demonstrated superior performance over the average-rank baseline, achieving higher sensitivity and specificity in identifying true CCIs.
• Application of CIS to scRNA-seq data from the intestine, skin, and uterus revealed both conserved and tissue-specific CCIs.
• Identified key conserved interactions (e.g., MIF-CD74, APP-CD74) and highlighted tissue-specific crosstalk, such as GUCA2A/GUCA2B-GUCY2C in the intestine and SPP1-PTGER4 in the uterus.

Conclusions:

• CIS offers a generalizable and effective framework for prioritizing CCIs from scRNA-seq data, surpassing naive consensus methods.
• The study establishes a valuable resource for distinguishing conserved and tissue-specific communication networks in epithelial barriers.
• Findings provide novel insights into barrier tissue function and pre-menopausal biology.