High-performance single-cell gene regulatory network inference at scale: the Inferelator 3.0

Abstract

Motivation Gene regulatory networks define regulatory relationships between transcription factors and target genes within a biological system, and reconstructing them is essential for understanding cellular growth and function. Methods for inferring and reconstructing networks from genomics data have evolved rapidly over the last decade in response to advances in sequencing technology and machine learning. The scale of data collection has increased dramatically; the largest genome-wide gene expression datasets have grown from thousands of measurements to millions of single cells, and new technologies are on the horizon to increase to tens of millions of cells and above. Results In this work, we present the Inferelator 3.0, which has been significantly updated to integrate data from distinct cell types to learn context-specific regulatory networks and aggregate them into a shared regulatory network, while retaining the functionality of the previous versions. The Inferelator is able to integrate the largest single-cell datasets and learn cell-type specific gene regulatory networks. Compared to other network inference methods, the Inferelator learns new and informative Saccharomyces cerevisiae networks from single-cell gene expression data, measured by recovery of a known gold standard. We demonstrate its scaling capabilities by learning networks for multiple distinct neuronal and glial cell types in the developing Mus musculus brain at E18 from a large (1.3 million) single-cell gene expression dataset with paired single-cell chromatin accessibility data. Availability The inferelator software is available on GitHub (https://github.com/flatironinstitute/inferelator) under the MIT license and has been released as python packages with associated documentation (https://inferelator.readthedocs.io/). Supplementary information Supplementary data are available at Bioinformatics online.

Coordinated control of neuronal differentiation and wiring by sustained transcription factors

“…We implemented the Inferelator 3.0 framework ( 40 ) to build gene regulatory network (GRN) models with the goal of gaining a more complete understanding of the regulatory programs used by developing optic lobe neurons. A key feature of this method is its use of transcription factor activity (TFA) that allows Inferelator to estimate the underlying activity of each TF using the expression levels of its known targets from prior information (“priors”) (Fig.…”

Coordinated control of neuronal differentiation and wiring by a sustained code of transcription factors

“…We implemented the Inferelator 3.0 framework 58 to build gene regulatory network (GRN) models with the goal of gaining a more complete understanding of the regulatory programs employed by developing optic lobe neurons. A key feature of this method is its use of Transcription Factor Activity (TFA) that allows Inferelator to estimate the underlying activity of each TF using the expression levels of its known targets from prior information (‘priors’, Fig.…”

“…Prior connectivity matrices (tables of 1s and 0s linking TFs to target genes) were determined separately for Tm and Lamina neurons using the P48 scATAC-seq data ( Fig. S4C ), Signac v1.3.0 and the Inferelator-Prior software 58 available on GitHub, from the release branch v0.2.3 faf5e47.…”

“…The networks were inferred using the Inferelator 3.0 58 , GitHub version v0.5.6 dd532f4, branch: release. Networks were learned using the multitask framework “AMuSR” 88 with the following parameters: regression=”amusr”, workflow=”multitask”, bootstraps=5, and cross validation seed set to 42 .…”

“…As part of the inference step, a linear model for each task is constructed that assumes each gene in the expression dataset can be modeled as a linear combination of the TFs regulating it ( X = βA ). Task specific betas represent the variance explained by each regulatory TF and are solved in the following way: We calculate the ratio between the variance of the residuals in the full model and the variance of the residuals when the model is refit, leaving one TF out at a time, as originally explained in the Inferelator 58 . Any non-zero beta is considered evidence for a regulatory relationship between a gene and a TF.…”