Gene regulatory networks. Methods and protocols (Q1721981): Difference between revisions

This book comprises of sixteen chapters that tackle different angles and methods proposed for the inference and characterization of regulatory networks. This heterogeneous collection of approaches is a timely addition to a fast progressing field and presents itself both as a comprehensive introductory overview of the field and as a stepping stone for future advances. The book commences with an overarching overview of the computational aspects related to the inference of gene regulatory networks provided by the editors. The introduction of the biological problem and its dependencies with the high throughput approaches, currently widely available and intensively used, are followed by the mathematical formulation and the description of few data driven, probabilistic and dynamical approaches. Evaluation methods and a list of tools are also included. The next two chapters focus on time series; in the second one, the authors present a method for inferring networks based on statistical inference using dynamic Bayesian networks (DBNs); extensions to nonlinear time dynamics and the consequences of missing variable are also discussed. In the third chapter, the authors give a broad overview of the state-of-the-art, recent methods for inferring networks from time series data, providing guidance and practical pointers for publicly available software. The next four chapters discuss different sides of causal networks, applied at genomic scale. In the fourth chapter, the authors use whole transcriptomic datasets (genomic and transcriptomic data) to infer such networks; the examples are built on 3000 genes from the Geuvadis dataset using Findr. The causal analysis is continued in the fifth chapter in which the authors present an empirical study in small networks, using an underlying Bayesian approach. The next chapter uses a breast cancer setup to illustrate a multi-attribute Gaussian graphical model (GGM) for inferring multiscale networks; a group-Lasso approach is used for linking the proteomic and the transcriptomic levels. In the seventh chapter, the authors present integrative approaches for the inference of networks at genome scale; two classes of methods are discussed in detail: (i) integrative approaches for linking transcription factors to their target genes and (ii) methods to link upstream PPI networks to proteomic datasets. The next two chapters use tree-based methods; in the eighth chapter, the authors describe an approach for the inference of regulatory networks using decision trees and random forests (GENIE3) targeted at detecting multivariate interaction effects. The following chapter focuses on a tree-based method for inferring topologies and dynamics, implemented in Jump3; this tool is a hybrid approach that combines the features of model-free and model-based methods using stochastic differential equations to model the gene expressions. The tenth chapter present methods at a different scale; the inference is performed from single cell data, that facilitate the identification of more complex, nonlinear dependencies between genes through the harnessing of the number of cells being sampled that aid the deconvolution of the true contribution of genes to the overall network. The next three chapters tackle the multiple sources of variation; in the eleventh chapter, the authors present approaches for inferring networks from multiple datasets using Gaussian process dynamical systems (GPDS); the features of these approaches are illustrated with examples, also made available as Jupyter notebooks. The twelfth chapter focuses on ensemble methods (based on the identification of a consensus signal) for the unsupervised identification of GRNs; the new method ScaleLSum is discussed in detail. In the thirteenth chapter, the authors discuss approaches for learning differential module networks across multiple experimental conditions; the examples are based on the Lemon-Tree software and human datasets. \par In the next chapter, the authors assess the stability of GRN inference, using subsampling and assessing the robustness of algorithms and the effect of the noise in the data. The main focus is on a set of parameters, NetSI which provide statistics of distances between the graphs (networks) generated on different runs using the Hamming-Ipsen-Mikhailov distance. In the fifteenth chapter, the authors evaluate the impact of the inference of gene regulatory networks for the understanding of biological processes and for the statistical modelling involved in designing the experimental setup of future experiments; the analysis is performed comprehensively from mRNA transcription to protein decay, the impact of topological properties of the inferred networks is also presented. The book concludes with a chapter on the scalability of ordinary differential equations models for biochemical processes, with an emphasis on the inference and effects of model parameters. The book covers a variety of computational and mathematical aspects related to the inference of gene regulatory networks. The style of the chapters seamlessly binds the comprehensive overview that recommended the book to junior researchers and the thorough description of topics, highlighting new direction of research, that would appeal to post-graduates and established researchers.