Network-Based Integrative Analysis of Genomics, Epigenomics and Transcriptomics in Autism Spectrum Disorders

Abstract

Current studies suggest that autism spectrum disorders (ASDs) may be caused by many genetic factors. In fact, collectively considering multiple studies aimed at characterizing the basic pathophysiology of ASDs, a large number of genes has been proposed. Addressing the problem of molecular data interpretation using gene networks helps to explain genetic heterogeneity in terms of shared pathways. Besides, the integrative analysis of multiple omics has emerged as an approach to provide a more comprehensive view of a disease. In this work, we carry out a network-based meta-analysis of the genes reported as associated with ASDs by studies that involved genomics, epigenomics, and transcriptomics. Collectively, our analysis provides a prioritization of the large number of genes proposed to be associated with ASDs, based on genes’ relevance within the intracellular circuits, the strength of the supporting evidence of association with ASDs, and the number of different molecular alterations affecting genes. We discuss the presence of the prioritized genes in the SFARI (Simons Foundation Autism Research Initiative) database and in gene networks associated with ASDs by other investigations. Lastly, we provide the full results of our analyses to encourage further studies on common targets amenable to therapy.

Network Structure Analysis Identifying Key Genes of Autism and Its Mechanism

“…Nanni et al [15] pointed out that while the analysis of epigenomics and transcriptomics from brain-derived samples can provide important insights into the potential mechanisms of disease etiology, there are relevant limitations with these types of studies (e.g., the quality of autopsy-derived tissue, sample size, influence of life experience, and cause of death) [22]. ese barriers have been overcome by analyzing blood samples, and recent blood-based works have shown the usefulness of this alternative approach to gather insights into autism [23][24][25].…”

“…Obviously, these data are the basis of systematic study of biology. Nanni et al [15] made a statement that large genome-wide association studies (GWAS), Copy Number Variation (CNV) testing, and genome sequencing yielded many nonoverlapping genes, a fact that underlines the complex genetic heterogeneity of autism [16] and reflects the architecture of intracellular networks, in which several possible combinations of genetic variations are likely to lead to a common pathological phenotype [17,18]. So, networkbased analysis will be helpful to study the pathogenesis of autism.…”

“…So, networkbased analysis will be helpful to study the pathogenesis of autism. Nanni et al [15] pointed out that one of the challenges that network-based analyses face is the identification of the so-called disease modules, that is, gene networks associated with diseases [17]. Hu et al [19] reported that stratifying the sample by cluster analyses revealed quantitative differences in gene expression which appear to correlate with severity of autism phenotype as well as gene expression profiles for each subtype which associate a "biological phenotype" (i.e., gene expression profile) to the respective functional/behavioral phenotype.…”

“…Currently, the integrative analysis of multiple omics has emerged as an approach to provide a more comprehensive view of a disease. Nanni et al [15] carried out a network-based meta-analysis of the genes reported as associated with autism by studies that involved genomics, epigenomics, and transcriptomics. e structure of a network determines the function of the network system.…”

SFARI Genes and where to find them; classification modelling to identify genes associated with Autism Spectrum Disorder from RNA-seq data

“…Despite the differences between the SFARI-gene curated list and transcriptomic data, they are frequently used together, either using genes from SFARI-gene to design transcriptomic experiments or to validate results (Araujo et al ., 2017; Berto et al ., 2018; Gokoolparsadh et al ., 2017; Lombardo et al ., 2017; Nowakowski et al ., 2017; Yu and He, 2017; Suetterlin et al ., 2018; Wang et al ., 2018). More recently, studies have begun to combine information from these two sources into single models that learn jointly from these data (Brueggeman et al ., 2020; Cogill and Wang, 2016; Di Nanni et al ., 2019; Lin et al ., 2020).…”