TY - JOUR T1 - Aberrant gene expression in autism JF - bioRxiv DO - 10.1101/029488 SP - 029488 AU - Jinting Guan AU - Ence Yang AU - Jizhou Yang AU - Gang Wang AU - Yong Zeng AU - Guoli Ji AU - James J. Cai Y1 - 2015/01/01 UR - http://biorxiv.org/content/early/2015/10/19/029488.abstract N2 - Autism spectrum disorder (ASD) is a complex neurodevelopmental condition characterized by phenotypic and genetic heterogeneity. Analyzing gene expression in autistic brains may reveal specific patterns of gene activation and repression associated with the disorder. For this, we developed a new analysis method called aberrant gene expression analysis, based on a multivariate distance measure for outlier detection, to identify aberrantly expressed genes. In a two-group setting, our method detects the discrepancies in gene expression dispersion between populations. Using this method, we re-visited RNA sequencing data previously generated from post-mortem brain tissues of 47 autistic and 57 control samples and identified a number of sets of genes with increased expression variability in autistics. In other words, the expression dispersion of these genes in autistics is more pronounced than that in controls. Many of those genes such as those expressed in synapse and those involved in neuropeptide binding are known to be implicated in ASD. We also found that many co-expressed gene modules present among non-autistic controls disappear among autistics, due to the aberrant gene expression that exclusively affects ASD patients. For the diagnostic purposes, we used a greedy algorithm to globally search for classifier gene sets, for which the expression in peripheral blood samples of autistics is maximally deviant from that of non-autistic controls. The aberrant gene expression respecting the classifier gene sets is more pronounced and specific for ASD samples, allowing the distinguishing of ASD from non-ASD samples. These results suggest that aberrant gene expression has the potential to be used as biomarkers for ASD. Taken together, we have developed a new gene expression analysis method based on a multivariate, dispersion-specific measure, which is powerful in detecting increased gene expression variability functionally associated with ASD. We conclude that detecting aberrant gene expression and identifying the underlying genes represent a new discovery and diagnostic strategy for ASD and other genetically heterogeneous disorders.Author Summary There is substantial phenotypic and genetic heterogeneity in autism, which complicates studies seeking to identify genetic factors that contribute to the disorder. However, despite this complexity, study designs have focused on using group differences, e.g., in gene expression, between autistic cases and controls to identify the genetic effects. The problem is that, by their nature, group difference-based methods, such as differential expression (DE) analysis, blur or collapse the heterogeneity within autistics, which, in fact, characterizes this spectrum disorder. For instance, DE analysis identifies genes expressed differentially between groups, but, by its design, the method tends to identify genes expressed uniformly or less variably across individuals within groups. By ignoring genes with variable expression in autistic individuals, an important axis of genetic heterogeneity contributing to gene expression variability among affected individuals has been overlooked. We have developed a new analysis method called aberrant gene expression analysis, aimed to identify genes with significant changes in expression variance between diseased and non-diseased samples. This is in sharp contrast to the purpose of conventional DE analysis method, which aims to identify genes with significant changes in expression mean. Using this new method, we detected increased gene expression variability and identified candidate genes with functions relevant to autism. ER -