Blog

Roser Vento, Wellcome Sanger Institute, designed CellPhoneDB, a novel repository of cells and their interactions, and applied this computational and genomics tool to study cellular connections from transcriptomics data. In this blog, Roser explains why CellPhoneDB is a unique database and why her team encourages fellow researchers to share their own cell interactions on the database.

Since my undergraduate, I have always been passionate about genomics. During my PhD in the Ballestar lab, Barcelona, I used genomics tools to characterise the signalling cascade that determines cell fate decisions in the myeloid lineage during inflammation. In 2016, I moved to Cambridge as an EMBO/HFSP postdoctoral fellow in the Teichmann lab, where I used single-cell technologies to gain insights into the influence of the microenvironment in cell identity, differentiation and fate. By then, droplet-based single cell RNA sequencing, allowed us to analyse thousands of cells in parallel, which meant that we could quantify the cellular composition of the complex human tissues.

Driven by my interest in cell-cell communication, I developed CellPhoneDB, a computational tool that enables users to assign potential interacting partners from single-cell transcriptomics data. I applied this tool to disentangle the complex communication network between the mother and the fetus during pregnancy. I started as a group leader at the Wellcome Sanger Institute in 2019.

Our group continues developing CellPhoneDB and adapting the tool to the new genomics technologies, including spatial transcriptomics. This computational tool is key to the other projects in the lab focused on the use of computational and genomics tools to deconstruct the molecular mechanisms underpinning cellular identity in human development and inflammation.

What makes CellPhoneDB unique?

Cells are not isolated entities, instead, they are continuously exposed to the environmental cues encountered in a tissue. Binding of a signal molecule present in the tissue (ligand) to its receiving molecule expressed in a specific cell (receptor) triggers sophisticated intracellular signalling pathways that influences cellular fate and function.

CellPhoneDB is a database of ligands, receptors and its interactions. The uniqueness of CellPhoneDB is that it considers the heteromeric composition of the ligands and receptors. This is important as the function of a receptor is determined by the expression and assembly of multiple subunits. In fact, it is common that receptors with antagonist function shares subunits – eg. interleukin family.

In addition, CellPhoneDB is integrated with a statistical tool that quantifies the expression of ligands and receptors on the distinct cell types identified by single-cell transcriptomics data and selects those interactions that are unique between cell types. Specific interactions within cells can inform us about the unique function and regulation of cell types.

A portal for researchers to contribute to interactions

CellPhoneDB is a curated database. In other words, only interactions that are supported by a bibliography are included in our resource. Attempts for a more systematic approach have failed in our hands, as they introduce lots of false-positive interactions which generate noise in the communication network.

We have curated thousands of interactions, but we encourage experts in the field to share their own interactions with us. This will facilitate an increase of the interactions stored in our database, which will ultimately influence the quality of the interacting networks that the users obtain from querying their single-cell transcriptomics data.

A valuable resource

CellPhoneDB is one of the most utilised tools to infer cell-cell communication pairs from single-cell transcriptomics data. In only two  years, our resource has more than 300 citations, and the webserver is interrogated by more than 500 users a month. In the original work, we used CellPhoneDB to study the communication between the mother and the fetus at the human placental-decidual interface during the early stages of pregnancy. Here, we showed that the dialogue between the maternal decidual natural killer cells (dNK) and the fetal trophoblast, enhances the homeostatic roles of dNK while at the same time downplays any potential killing function of dNKs.

In addition, CellPhoneDB has been used in other human tissues to characterise the interactions underpinning disease conditions, such as asthma or liver cirrhosis. In collaboration with the Ballestar lab, our team has recently used CellPhoneDB to infer the aberrant immune interactions underpinning common variable immunodeficiency (manuscript under submission). Here, we have observed an altered communication between B and T cells, which has consequences for B cell maturation and function. Application of CellPhoneDB to a disease context is highly relevant as receptors are surface proteins that can be targeted by drugs.

To advance knowledge and explore new angles

Sharing data empowers collaboration between researchers and this is crucial to advance our scientific knowledge. Genomics data contains loads of information which means that one can find novelty in the results by analysing it in multiple ways. Similarly, people with a distinct scientific background may find unique angles when analysing the same dataset, which may have gone unnoticed by others. For example, our study in the maternal-fetal interface has recently been interrogated for the SARS-CoV-2 entry pathways. These studies have observed that fetal trophoblast in the placenta express high levels of ACE2 while the protease TMPRSS2 is lowly expressed.

In our team, we like to share the data using friendly, open-source platforms which allows researchers with no genomic background to navigate throughout our resources. In addition to data sharing, I think it is also relevant to share the experimental protocols and code to enhance data reproducibility. There are platforms available for that purpose, such as protocols.io (experimental protocols) or github (code), which we utilise regularly in our teams.