Ingres : from single-cell RNA-seq data to single-cell probabilistic Boolean networks

Abstract Motivation The current explosion of ’omics data has provided scientists with an unique opportunity to elucidate the inner workings of biological processes that remained opaque. For this, computational models are essential. Gene regulatory networks (GRN) have long been used as a way to integrate heterogeneous data into a discrete model, and are very useful to generate actionable hypotheses on the mechanisms governing these biological processes. Boolean networks are particularly popular for this kind of discrete models. When working with single-cell RNA-seq datasets a main focus of analysis is the differential expression between subpopulations of cells. Boolean networks are limited in this task, since they cannot easily represent different levels of expression, confined as they are to binary states. We set out to develop an algorithm that can fit Boolean networks with this kind of data, maintaining both the heterogeneity of the data and the simplicity and computational efficiency of these type of networks. Results Here we present Ingres ( I nferring Probabilistic Boolean N etworks of G ene R egulation Using Protein Activity E nrichment S cores) an open-source tool that uses single-cell sequencing data and prior knowledge GRNs to produce a probabilistic Boolean network (PBN) per each cell and/or cluster of cells in the dataset. Ingres allows to better capture the differences between cell phenotypes, using a continuous measure of protein activity while still confined to the simplicity of a GRN. We believe Ingres will be useful to better understand the heterogeneous makeup of cell populations, to gain insight into the specific circuits that drive certain phenotypes, and to use expression and other omics to infer computational cellular models in bulk or single-cell data. Availability and implementation Ingres has been implemented as an R package, and it is publicly available at https://github.com/CBigOxf/ingres . It is currently being submitted to the public repository CRAN too. works seamlessly with existing software for single-cell RNA-seq analysis, and for network analysis, modelling and visualization. Contact pedro.victori@oncology.ox.ac.uk ; francesca.buffa@unibocconi.it ; francesca.buffa@imm.ox.ac.uk Supplementary information Supplementary data are available online. Software documentation and an explanatory vignette are available on GitHub and as part of the R package.