dynDeepDRIM: a dynamic deep learning model to infer direct regulatory interactions using single cell time-course gene expression data

Abstract

Time-course single-cell RNA sequencing (scRNA-seq) data have been widely applied to reconstruct the cell-type-specific gene regulatory networks by exploring the dynamic changes of gene expression between transcription factors (TFs) and their target genes. The existing algorithms were commonly designed to analyze bulk gene expression data and could not deal with the dropouts and cell heterogeneity in scRNA-seq data. In this paper, we developed dynDeepDRIM that represents gene pair joint expression as images and considers the neighborhood context to eliminate the transitive interactions. dynDeepDRIM integrated the primary image, neighbor images with time-course into a four-dimensional tensor and trained a convolutional neural network to predict the direct regulatory interactions between TFs and genes. We evaluated the performance of dynDeepDRIM on five time-course gene expression datasets. dynDeepDRIM outperformed the state-of-the-art methods for predicting TF-gene direct interactions and gene functions. We also observed gene functions could be better performed if more neighbor images were involved.

Depicting pseudotime-lagged causality across single-cell trajectories for accurate gene-regulatory inference

“…Building upon several deep learning-based methods ( 18 , 27 ), we have demonstrated that combining fully supervised deep learning with joint probability matrices of pseudotime-lagged single-cell trajectories can overcome certain limitations of current Granger causality-based methods of gene-regulatory inference ( 9 , 12 ), such as their inability to infer cyclic regulatory motifs. Although our convolutional neural network DELAY remains sensitive to the ordering of single cells in pseudotime, the network can nevertheless accurately infer direct gene-regulatory interactions from both time course and snapshot data sets, unlike many supervised methods that rely strongly on the number of available time course samples ( 15 , 16 ). We suspect that DELAY is sensitive to the specific ordering of adjacent cells in trajectories because pseudotime inference methods such as Slingshot ( 6 ) infer lineages from minimum spanning trees that directly depend on cell-to-cell similarities in gene expression values ( 7 ).…”

“…By contrast, deep learning-based methods make no assumptions about the temporal relationships or connectivity between genes in complex regulatory networks; instead, these data-driven approaches learn general features of regulatory interactions ( 15 , 16 ). Here, we describe a deep learning-based method termed De picting La gged Causalit y (DELAY) that learns gene-regulatory interactions from discrete joint probability matrices of paired, pseudotime-lagged gene expression trajectories.…”

Depicting pseudotime-lagged causality across single-cell trajectories for accurate gene-regulatory inference

“…Building upon several deep learning-based methods 18, 22 , we have demonstrated that combining fully-supervised deep learning with joint-probability matrices of pseudotime-lagged single-cell trajectories can overcome certain limitations of current Granger causality-based methods of gene-regulatory inference 9, 12 , such as their inability to infer cyclic regulatory motifs. Moreover, although our convolutional neural network DELAY remains sensitive to the ordering of single cells in pseudotime, the network can accurately infer direct gene-regulatory interactions from both timecourse and snapshot datasets, unlike many supervised methods that rely strongly on the number of available time-course samples 15, 16 . We suspect that DELAY is sensitive to the specific ordering of adjacent cells in trajectories because pseudotime inference methods such as Slingshot 6 infer lineages from minimum spanning trees that directly depend on cell-to-cell similarities in gene-expression values 7 .…”

“…By contrast, deep learning-based methods make no assumptions about the temporal relationships or connectivity between genes in complex regulatory networks; instead, these data-driven approaches learn general features of regulatory interactions 15, 16 . Here we describe a deep learning-based method termed DELAY ( De picting La gged Causalit y ) that learns gene-regulatory interactions from discrete joint-probability matrices of paired, pseudotime-lagged gene-expression trajectories.…”