Predicting cell-to-cell communication networks using NATMI.

Hou R, Denisenko E, Ong HT, Ramilowski JA, Forrest ARR. · 2020

Read Predicting cell-to-cell communication networks using NATMI. by Hou R, Denisenko E, Ong HT, Ramilowski JA, Forrest ARR. (2020) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/europmc-pmc7538930.txt.

Introduction

Hou and co-authors present NATMI, a toolkit for predicting how cells signal each other. Single-cell sequencing now profiles thousands of cells at once. Knowing which cells talk helps explain tissue behavior.

NATMI draws on a database of 2293 ligand-receptor pairs. Each pair has literature support behind it. The tool builds and shows communication networks from expression data. Tests on published datasets found the busiest cell pairs. They also found the most active signaling pairs. Results pointed to strong self-signaling in many cells. The toolkit works with single-cell and bulk data. It also compares networks across conditions.

Catalog record

CREATORHou R, Denisenko E, Ong HT, Ramilowski JA, Forrest ARR.
YEAR2020
COLLECTIONPAPERS
FORMATTXT
FILE SIZE66 KB
LICENSECC BY 4.0
RIGHTSCC BY
SOURCEEurope PMC

Source: https://europepmc.org/articles/PMC7538930

Browse the full library · About the Tower of Babel

Full text

The opening of the work, from the library's hosted text:

Predicting cell-to-cell communication networks using NATMI By Rui Hou, Elena Denisenko, Huan Ting Ong, Jordan A Ramilowski, Alistair R R Forrest Published in: Nature Communications DOI: 10.1038/s41467-020-18873-z ABSTRACT Abstract Development of high throughput single-cell sequencing technologies has made it cost-effective to profile thousands of cells from diverse samples containing multiple cell types. To study how these different cell types work together, here we develop NATMI (Network Analysis Toolkit for Multicellular Interactions). NATMI uses connectomeDB2020 (a database of 2293 manually curated ligand-receptor pairs with literature support) to predict and visualise cell-to-cell communication networks from single-cell (or bulk) expression data. Using multiple published single-cell datasets we demonstrate how NATMI can be used to identify (i) the cell-type pairs that are communicating the most (or most specifically) within a network, (ii) the most active (or specific) ligand-receptor pairs active within a network, (iii) putative highly-communicating cellular communities and (iv) differences in intercellular communication when profiling given cell types under different conditions. Furthermore, analysis of the Tabula Muris (organism-wide) atlas confirms our previous prediction that autocrine signalling is a major feature of cell-to-cell communication networks, while also revealing that hundreds of ligands and their cognate receptors are co-expressed in individual cells suggesting a substantial potential for self-signalling. Subject terms: Cellular signalling networks, Multicellular systems, Regulatory networks INTRODUCTION In 2015, we published the first draft map of predicted cell-to-cell communication between major human cell types based on the expression levels of 708 ligands and 691 receptors measured in 144 purified human primary cell types, and a manually curated set of 2422 human ligand–receptor pairs (1894 pairs had literature support)1. Our major findings were that any given cell type expresses tens to hundreds of different ligands and receptors, that extensive autocrine signalling is a common feature of all cell types studied thus far and that cell-to-cell communication networks are highly connected via hundreds of predicted ligand–receptor paths. Recently, with the availability of high throughput single-cell platforms2–10, expression profiling across an ever-growing variety of cell types from dissociated tissues without prior purification has become common. To understand how individual cells and cell types within these complex samples communicate, multiple groups have since used our ligand–receptor pairs from 20151 to infer cell-to-cell communication in developing heart11, kidney12, liver13, lung14, cancer15,16, and cortex17. Here, we present NATMI (Network Analysis Toolkit for Multicellular Interactions), a user-friendly tool primarily designed to process single-cell gene expression datasets, which can also be applied to bulk transcriptomics and proteomics data. In brief, NATMI uses connectomeDB2020 (a newly updated curated ligand–receptor pair list) or user-specified ligand–receptor pairs to predict and visualise the network of cell-to-cell communication between cell types (or clusters) in these datasets (Supplementary Fig. 1). We demonstrate several ways NATMI can be used to summarise and extract biological insights from cell-to-cell communication networks (Supplementary Fig. 1). Specifically, NATMI can: (1) show all cell types predicted to communicate via a user-specified ligand–receptor pair (Supplementary Fig. 1d); (2) show all ligand–receptor pairs used for communication between a user-specified pair of cell types (Supplementary Fig. 1e); (3) summarise the entire communication network to show how strongly or specifically each cell type in a complex sample communicates to every other cell type, thus identifying highly communicating cell pairs or communities (Supplementary Fig. 1f); and (4) compare communication networks from two different conditions and identify edges (ligand–receptor pairs) that differ (delta network) between them (Supplementary Fig. 1g). NATMI (Python script) and the updated ligand–receptor lists are freely available at https://github.com/forrest-lab/NATMI/. RESULTS Updated ligand–receptor pair lists To facilitate the exploration of intercellular interactions, in 2015 we published a set of 1894 ligand–receptor pairs with primary literature support and an additional 528 putative pairs (secreted and plasma-membrane proteins with high throughput protein–protein interaction (PPI) evidence)1. Here, we present connectomeDB2020, an updated set of 2293 ligand–receptor pairs with primary literature support (Supplementary Data 1). These consist of 1751 pairs from our 2015 resource, 121 pairs from CellphoneDB v2.018, 50 pairs from RNA-Magnet19, 22 pairs from SingleCellSignalR20, 9 pairs from ICELLNET21, and 340 new manually curated pairs predominantly reported since the original publication. Supplementary Data 1 also lists pairs from each resource excluded due to lack of primary evidence (these include 143 of our original pairs removed after user feedback and further checks revealed that the PubMed ID supplied from HPRD22 was incorrect and did not support the interaction). To allow users to discriminate contact-dependent signalling from soluble ligand-mediated signalling, we have now annotated all ligands as either secreted, plasma membrane or both (Methods). Additionally, to facilitate the use of NATMI for other organisms, we provide inferred ligand–receptor pairs for multiple vertebrate species based on human ortholog mappings provided in the NCBI HomoloGene Database23. Predicting cell-to-cell communication using NATMI The workflow for cell-to-cell communication analysis in NATMI starts by the user providing input data files of gene expression and cell labels for single-cell data analysis. Using the connectomeDB2020 ligand–receptor pair list described above, NATMI extracts the expression levels of every ligand and receptor expressed in each cell type (Supplementary Fig. 1a–c). Edges between any pair of cell types are then predicted based on the expression of the ligand in one cell type and the expression of its cognate receptor in the other cell type (Supplementary Fig. 2a). As cells can express multiple ligands and cognate receptors, cell pairs are connected by multiple edges that are accordingly weighted by the expression of ligands and receptors in these cell types (Supplementary Fig. 2b). Moreover, the same ligands and receptors can be expressed by multiple cell types. This makes any given ligand–receptor pair a hyperedge that can connect multiple cell types and the underlying structure of a real cell-to-cell communication network—a weighted-directed-multi-hyperedge network (Supplementary Fig. 2c). In NATMI, however, we reduce these to weighted-directed-multi-edge networks (Supplementary Fig. 2d) and for simplicity refer to these as ‘cell-to-cell communication networks’. Lastly, NATMI introduces the concept of cell-connectivity-summary networks that merge the many ligand–receptor edges drawn from one cell type to another into a combined weighted cell-connectivity-summary edge to summarise how strongly (or specifically) each cell type is communicating to another cell type (Supplementary Fig. 1f). Edge weights in cell-to-cell communication networks For each analysed dataset NATMI creates an edge file that summarises the levels and fractions of cells in each cell type expressing each ligand and receptor. From this it calculates two different edge weights. The mean-expression edge weights are calculated by multiplying the mean-expression level of the ligand in the sending cell type by the mean expression of the receptor in the target cell type. This weighting is useful to emphasise highly expressed ligands and receptors but provides no discrimination between cell-type-specific and housekeeping edges. The specificity-based edge weights, on the other hand, help identify the most specific edges in the network regardless of expression levels and are calculated as the product of the ligand and receptor specificities, where each specificity is defined as the mean expression of the ligand/receptor in a given cell type divided by the sum of the mean expression of that ligand/receptor across all cell types. The specificity-based edge weights range from 0 to 1 where a weight of 1 means both the ligand and receptor are only expressed in one (not necessarily the same) cell type. To better demonstrate the utility

Read the complete text