Deciphering gene regulation from gene expression dynamics using deep neural network

Abstract

Complex biological functions are carried out by the interaction of genes and proteins.Uncovering the gene regulation network behind a function is one of the central themes in biology.Typically, it involves extensive experiments of genetics, biochemistry and molecular biology. In this paper, we show that much of the inference task can be accomplished by a deep neural network (DNN), a form of machine learning or artificial intelligence. Specifically, the DNN learns from the dynamics of the gene expression. The learnt DNN behaves like an accurate simulator of the system, on which one can perform in-silico experiments to reveal the underlying gene network. We demonstrate the method with two examples: biochemical adaptation and the gap-gene patterning in fruit fly embryogenesis. In the first example, the DNN can successfully find the two basic network motifs for adaptation -the negative feedback and the incoherent feed-forward. In the second and much more complex example, the DNN can accurately predict behaviors of essentially all the mutants. Furthermore, the regulation network it uncovers is strikingly similar to the one inferred from experiments. In doing so, we develop methods for deciphering the gene regulation network hidden in the DNN "black box". Our interpretable DNN approach should have broad applications in genotype-phenotype mapping. SignificanceComplex biological functions are carried out by gene regulation networks. The mapping between gene network and function is a central theme in biology. The task usually involves extensive experiments with perturbations to the system (e.g. gene deletion). Here, we demonstrate that machine learning, or deep neural network (DNN), can help reveal the underlying gene regulation for a given function or phenotype with minimal perturbation data. Specifically, after training with wild-type gene expression dynamics data and a few mutant snapshots, the DNN learns to behave like an accurate simulator for the genetic system, which can be used to predict other mutants' behaviors. Furthermore, our DNN approach is biochemically interpretable, which helps 3 uncover possible gene regulatory mechanisms underlying the observed phenotypic behaviors. IntroductionComplex biological functions are carried out by gene regulation networks. Uncovering the gene regulation network behind a function is one of the central themes in biology. Traditionally, this task usually involves extensive genetic and biochemical experiments with perturbations to the biological system. For example, in the classical gene knockout experiments, by observing the expression of gene a increasing (decreasing) when deleting gene b, one may infer that gene a is repressed (activated) by gene b. More recently, statistical and bioinformatical methods have been used to help mapping out genetic and protein interactions. These methods can be very powerful especially in analyzing high throughput experimental data and extracting information about correlations among the genes and proteins (1, 2). Another computational approach is...

Fully interpretable deep learning model of transcriptional control

“…Deep learning has been widely deployed in genomics and systems biology over the last few years ( Alipanahi et al , 2015 ; Avsec et al , 2019 ; Celesti et al , 2017 ; Cuperus et al , 2017 ; Greenside et al , 2018 ; Koh et al , 2017 ; Libbrecht and Noble, 2015 ; Movva et al , 2019 ; Nair et al , 2019 ; Pouladi et al , 2015 ; Rui et al , 2007 ; Shen et al , 2018 ). Many of the developed tools have been highly successful in classification problems such as the identification of binding sites, open regions of chromatin and the location of enhancers.…”

Adaptation, fitness landscape learning and fast evolution

“…This network modifies the thresholds by a simple feedback mechanism. Although we are not aware of clear experimental evidence for the existence of such a mechanism, we nevertheless think that such a mechanism can be connected with regulations via enhancers 47, 48 , where enhancer action is described by deep network models based on thermodynamics, and chemical kinetics, and those models contain threshold parameters. Alternative variants involve modifications of weights.…”

The Cognitive Lens: a primer on conceptual tools for analysing information processing in developmental and regenerative morphogenesis

“…The central idea that connects neural processing and patterning is succinctly laid out by Grossberg [21]: it is the same principle at play in the brain, where neurons sense information, communicate it among themselves and construct neural patterns, that must underlie patterning in systems outside of the brain as well. Accordingly, ANNs constitute an important class of models in various domains of developmental biology: Planarian morphogenesis [91], Drosophila embryogenesis [92] and many more [93]. In all these models, ANNs either learn or control the corresponding developmental process.…”