Unraveling cell-cell communication with NicheNet by inferring active ligands from transcriptomics data

TL;DR

This paper presents a comprehensive workflow and best practices for using NicheNet to infer ligand-receptor interactions from single-cell transcriptomics data, enhancing user guidance and analysis efficiency.

Contribution

It introduces a best practices workflow for NicheNet, including new data updates, ligand prioritization methods, and user-friendly tools for diverse experimental designs.

Findings

Updated data sources in NicheNet v2

New downstream ligand-receptor prioritization procedure

Wrapper functions for simplified analysis

Abstract

Ligand-receptor interactions constitute a fundamental mechanism of cell-cell communication and signaling. NicheNet is a well-established computational tool that infers ligand-receptor interactions that potentially regulate gene expression changes in receiver cell populations. Whereas the original publication delves into the algorithm and validation, this paper describes a best practices workflow cultivated over four years of experience and user feedback. Starting from the input single-cell expression matrix, we describe a "sender-agnostic" approach which considers ligands from the entire microenvironment, and a "sender-focused" approach which only considers ligands from cell populations of interest. As output, users will obtain a list of prioritized ligands and their potential target genes, along with multiple visualizations. In NicheNet v2, we have updated the data sources and implemented a downstream procedure for prioritizing cell-type-specific ligand-receptor pairs. Although a standard NicheNet analysis takes less than 10 minutes to run, users often invest additional time in making decisions about the approach and parameters that best suit their biological question. This paper serves to aid in this decision-making process by describing the most appropriate workflow for common experimental designs like case-control and cell differentiation studies. Finally, in addition to the step-by-step description of the code, we also provide wrapper functions that enable the analysis to be run in one line of code, thus tailoring the workflow to users at all levels of computational proficiency.