CINS: Cell Interaction Network inference from Single cell expression data
The authors present a tool by which using scRNA-Seq data, interactions between different cell types and the genes involved in those interactions can be inferred. The tool is initially showcased utilising human lung scRNA-Seq data for healthy controls vs an Idiopathic Pulmonary Fibrosis (IPF) cohort. This is furthered by application to murine lung samples from old and young murine models. Several significant cellular interactions generated from either application in the paper are shown such as the presence of Macrophage to Ciliated cell dependency in the human control cohort which is not present in IPF patients.
The authors openly acknowledge some of the short falls of this methodology which include the inability to include self-edges in the Bayesian Network (BN) meaning autocrine relationships cannot be inferred. Additionally, the BN is dependent on both the number of total samples and how many cells were profiled per sample for more robust network generation, which may impede analysis if the sample number is a limiting factor.
When the tool itself is applied within the paper a general critique would be the validation approach of the BN, only interactions generated by the BN are validated (which could be strengthened with additional literature references) however it may be advisable to generate a list of known interactions from published literature before any network generation and validate these entirely, this could show if any key known interactions are missing.
This paper provides a useful readily applicable tool to enable immunologists and other researchers to further utilise scRNA-Seq datasets. The components of the pipeline may not be individually novel but this does not detract from the final tool itself. A key benefit to this tool is that it can be applied ad-hoc to already generated samples without the requirement for any specific sample processing to be applied (which is the case for some other transcriptomic tools investigating cell-cell interaction). However, some of the limitations of the tool inevitably impact its significance, additionally the utilisation of this tool itself may not be readily understood by wet-lab based researchers and this can detract from its application.
Overall, this paper provides a tool which can aid in further studies by generating useful cell to cell interaction information.
Reviewed by Jonathan Hamp as part of the cross-institutional journal club of the Immunology Institute of the Icahn School of Medicine, Mount Sinai and the Kennedy Institute of Rheumatology, University of Oxford.