High-throughput RNA-sequencing (RNA-seq) technologies provide an unprecedented opportunity to explore the individual transcriptome. Unmapped reads are a large and often overlooked output of standard RNA-seq analyses. Here, we present Read Origin Protocol (ROP), a tool for discovering the source of all reads originating from complex RNA molecules. We apply ROP to samples across 2630 individuals from 54 diverse human tissues. Our approach can account for 99.9% of 1 trillion reads of various read length. Additionally, we use ROP to investigate the functional mechanisms underlying connections between the immune system, microbiome, and disease. ROP is freely available at https://github.com/smangul1/rop/wiki .

MicroRNAs orchestrate brain functioning via interaction with microRNA recognition elements (MRE) on target transcripts. However, the global impact of potential competition on the microRNA pool between coding and non-coding brain transcripts that share MREs with them remains unexplored. Here we report that non-coding pseudogene transcripts carrying MREs (PSG) often show duplicated origin, evolutionary conservation and higher expression in human temporal lobe neurons than comparable duplicated MRE-deficient pseudogenes (PSG). PSG participate in neuronal RNA-induced silencing complexes (RISC), indicating functional involvement. Furthermore, downregulation cell culture experiments validated bidirectional co-regulation of PSG with MRE-sharing coding transcripts, frequently not their mother genes, and with targeted microRNAs; also, PSG single-nucleotide polymorphisms associated with schizophrenia, bipolar disorder and autism, suggesting interaction with mental diseases. Our findings indicate functional roles of duplicated PSG in brain development and cognition, supporting physiological impact of the reciprocal co-regulation of PSG with MRE-sharing coding transcripts in human brain neurons.

Genetic susceptibility to intellectual disability (ID), autism spectrum disorder (ASD), and schizophrenia (SCZ) often arises from mutations in the same genes, suggesting that they share common mechanisms. We studied genes with de novo mutations in the three disorders and genes implicated in SCZ by genome-wide association study (GWAS). Using biological annotations and brain gene expression, we show that mutation class explains enrichment patterns more than specific disorder. Genes with loss-of-function mutations and genes with missense mutations were associated with different pathways across disorders. Conversely, gene expression patterns were specific for each disorder. ID genes were preferentially expressed in the cortex; ASD genes were expressed in the fetal cortex, cerebellum, and striatum; and genes associated with SCZ were expressed in the adolescent cortex. Our study suggests that convergence across neuropsychiatric disorders stems from common pathways that are consistently vulnerable to genetic variations but that spatiotemporal activity of genes contributes to specific phenotypes.

Recent successes in genome-wide association studies (GWASs) make it possible to address important questions about the genetic architecture of complex traits, such as allele frequency and effect size. One lesser-known aspect of complex traits is the extent of allelic heterogeneity (AH) arising from multiple causal variants at a locus. We developed a computational method to infer the probability of AH and applied it to three GWASs and four expression quantitative trait loci (eQTL) datasets. We identified a total of 4,152 loci with strong evidence of AH. The proportion of all loci with identified AH is 4%-23% in eQTLs, 35% in GWASs of high-density lipoprotein (HDL), and 23% in GWASs of schizophrenia. For eQTLs, we observed a strong correlation between sample size and the proportion of loci with AH (R = 0.85, p = 2.2 × 10), indicating that statistical power prevents identification of AH in other loci. Understanding the extent of AH may guide the development of new methods for fine mapping and association mapping of complex traits.

The study of the genetics of gene expression is of considerable importance to understanding the nature of common, complex diseases. The most widely applied approach to identifying relationships between genetic variation and gene expression is the expression quantitative trait loci (eQTL) approach. Here, we increased the computational power of eQTL with an alternative and complementary approach based on analyzing allele specific expression (ASE). We designed a novel analytical method to identify cis-acting regulatory variants based on genome sequencing and measurements of ASE from RNA-sequencing (RNA-seq) data. We evaluated the power and resolution of our method using simulated data. We then applied the method to map regulatory variants affecting gene expression in lymphoblastoid cell lines (LCLs) from 77 unrelated northern and western European individuals (CEU), which were part of the HapMap project. A total of 2309 SNPs were identified as being associated with ASE patterns. The SNPs associated with ASE were enriched within promoter regions and were significantly more likely to signal strong evidence for a regulatory role. Finally, among the candidate regulatory SNPs, we identified 108 SNPs that were previously associated with human immune diseases. With further improvements in quantifying ASE from RNA-seq, the application of our method to other datasets is expected to accelerate our understanding of the biological basis of common diseases.

Monoallelic expression, including genomic imprinting, X-chromosome inactivation and random monoallelic expression of autosomal genes are epigenetic phenomena. Genes that are expressed in a monoallelic way may be more vulnerable to genetic or epigenetic mutations. Thus, comprehensive exploration of monoallelic expression in human brains may shed light on complex brain disorders. Autism-related disorders are known to be associated with imprinted genes on chromosome 15. However, it is not clear whether other imprinted regions or other types of monoallelic expression are associated with autism spectrum disorder (ASD). Here, we performed a genome-wide survey of allele expression imbalance (AEI) in the human brain using single-nucleotide polymorphisms (SNPs), in 18 individuals with ASD and 15 controls. Individuals with ASD had the most extreme number of monoallelic expressed SNPs in both the autosomes and the X chromosome. In two cases that were studied in detail, the monoallelic expression was confined to specific brain region or cell type. Using these data, we were also able to define the allelic expression status of known imprinted genes in the human brain and to identify abnormal imprinting in an individual with ASD. Lastly, we developed an analysis of individual-level expression, focusing on the difference of each individual from the mean. We found that individuals with ASD had more genes that were up- or down-regulated in an individual-specific manner. We also identified pathways perturbed in specific individuals. These results underline the heterogeneity in gene regulation in ASD, at the level of both allelic and total expression.

Though extensively characterized clinically, the causes of autism spectrum disorder (ASD) remain a mystery. ASD is known to have a strong genetic basis, but it is genetically very heterogeneous. Recent studies have estimated that de novo disruptive mutations in hundreds of genes may contribute to ASD. However, it is unclear how it is possible for mutations in so many different genes to contribute to ASD. Recent findings suggest that many of the mutations disrupt genes involved in transcription regulation that are expressed prenatally in the developing brain. De novo disruptive mutations are also more frequent in girls with ASD, despite the fact that ASD is more prevalent in boys. In this paper, we hypothesize that loss of robustness may contribute to ASD. Loss of phenotypic robustness may be caused by mutations that disrupt capacitors that operate in the developing brain. This may lead to the release of cryptic genetic variation that contributes to ASD. Reduced robustness is consistent with the observed variability in expressivity and incomplete penetrance. It is also consistent with the hypothesis that the development of the female brain is more robust, and it may explain the higher rate and severity of disruptive de novo mutations in girls with ASD.

MicroRNAs (miRNAs) can repress multiple targets, but how a single de-balanced interaction affects others remained unclear. We found that changing a single miRNA-target interaction can simultaneously affect multiple other miRNA-target interactions and modify physiological phenotype. We show that miR-608 targets acetylcholinesterase (AChE) and demonstrate weakened miR-608 interaction with the rs17228616 AChE allele having a single-nucleotide polymorphism (SNP) in the 3'-untranslated region (3'UTR). In cultured cells, this weakened interaction potentiated miR-608-mediated suppression of other targets, including CDC42 and interleukin-6 (IL6). Postmortem human cortices homozygote for the minor rs17228616 allele showed AChE elevation and CDC42/IL6 decreases compared with major allele homozygotes. Additionally, minor allele heterozygote and homozygote subjects showed reduced cortisol and elevated blood pressure, predicting risk of anxiety and hypertension. Parallel suppression of the conserved brain CDC42 activity by intracerebroventricular ML141 injection caused acute anxiety in mice. We demonstrate that SNPs in miRNA-binding regions could cause expanded downstream effects changing important biological pathways.

MicroRNAs (miRNAs) have presumably contributed to the emergence of the novel expression patterns, higher brain functions, and skills underlying human evolution. However, it is incompletely understood how new miRNAs have evolved in the human lineage because their initial emergence predictably entailed deleterious consequences due to their powerful multitarget effects. Here, we report genetic variation and conservation parameters for miRNAs and their predicted targets in the genomes of 1,092 humans and 58 additional organisms. We show that miRNAs were evolutionarily more conserved than their predicted binding sites, which were inversely subject to the accumulation of single-nucleotide variations over short evolutionary timescales. Moreover, the predictably "younger" human-specific miRNAs presented lower genetic variation than other miRNAs; their targets displayed higher genetic variation compared with other miRNA targets in diverse human populations; and neuronal miRNAs showed yet lower levels of genetic variation and were found to target more protein-coding genes than nonneuronal miRNAs. Furthermore, enrichment analysis indicated that targets of human-specific miRNAs primarily perform neuronal functions. Specifically, the genomic regions harboring the vertebrate-conserved neuronal miRNA-132 presented considerably higher conservation scores than those of its target genes throughout evolution, whereas both the recently evolved human miRNA-941 and its acquired targets showed relatively low conservation. Our findings demonstrate inversely correlated genetic variation around miRNAs and their targets, consistent with theories of coevolution of these elements and the predicted role attributed to miRNAs in recent human evolution.

On April 2013 experts in the field of autism from Italy and Israel convened in Jerusalem to discuss and finalize clinical recommendations for early diagnosis and intervention in Autism Spectrum Disorders (ASDs). In this paper, we summarize the results of this Italian-Israeli consensus conference. ASDs constitute a class of severe and heterogeneous neurodevelopmental conditions caused by atypical brain development beginning during early prenatal life, reflecting many genetic, neurobiological and environmental influences. The first clinical signs of ASDs begin to be evident in children between 12 and 18 months of age, often after a period of relatively typical postnatal development. Recent longitudinal studies reveal substantial diversity in developmental trajectories through childhood and adolescence. Some intervention approaches have been demonstrated to be effective in improving core symptoms of ASDs, even if the heterogeneity and developmental nature of the disorder make it implausible that only one specific treatment will be best for all children with ASDs. More randomized control trials (RCTs) on early intervention are needed to identify the most effective strategies and provide the most efficient allocation of resources during the critical early intervention time period. Future research should focus on linking biological phenotypes with specific genotypes, thus establishing a foundation for the development of diagnostic screening tools and individualization of treatments.

Colorectal cancer develops in a sequential, evolutionary process, leading to a heterogenic tumor. Comprehensive molecular studies of colorectal cancer have been previously performed; still, the process of carcinogenesis is not fully understood. We utilized gene expression patterns from 94 samples including normal, adenoma, and adenocarcinoma colon biopsies and performed a coexpression network analysis to determine gene expression trajectories of 8,000 genes across carcinogenesis. We found that the majority of gene expression changes occur in the transition from normal tissue to adenoma. The upregulated genes, known to be involved in cellular proliferation, included c-Myc along with its targets. In a cellular model system, we show that physiologic upregulation of c-Myc can lead to cellular proliferation without DNA replication stress. Our analysis also found that carcinogenesis involves a progressive downregulation of genes that are markers of colonic tissue and propose that this reflects a perturbed differentiation of colon cells during carcinogenesis. The analysis of miRNAs targets pointed toward the involvement of miR17 in the regulation of colon cell differentiation. Finally, we found that copy-number variations (CNV) enriched in colon adenocarcinoma tend to occur in genes whose expression changes already in adenoma, with deletions occurring in genes downregulated and duplications in genes upregulated in adenomas. We suggest that the CNVs are selected to reinforce changes in gene expression, rather than initiate them. Together, these findings shed new light into the molecular processes that underlie the transformation of colon tissue from normal to cancer and add a temporal context that has been hitherto lacking.

Pantothenate Kinase-Associated Neurodegeneration (PKAN) is a neurodegenerative disorder with a poorly understood molecular mechanism. It is caused by mutations in Pantothenate Kinase, the first enzyme in the Coenzyme A (CoA) biosynthetic pathway. Here, we developed a Drosophila model of PKAN (tim-fbl flies) that allows us to continuously monitor the modeled disease in the brain. In tim-fbl flies, downregulation of fumble, the Drosophila PanK homologue in the cells containing a circadian clock results in characteristic features of PKAN such as developmental lethality, hypersensitivity to oxidative stress, and diminished life span. Despite quasi-normal circadian transcriptional rhythms, tim-fbl flies display brain-specific aberrant circadian locomotor rhythms, and a unique transcriptional signature. Comparison with expression data from flies exposed to paraquat demonstrates that, as previously suggested, pathways others than oxidative stress are affected by PANK downregulation. Surprisingly we found a significant decrease in the expression of key components of the photoreceptor recycling pathways, which could lead to retinal degeneration, a hallmark of PKAN. Importantly, these defects are not accompanied by changes in structural components in eye genes suggesting that changes in gene expression in the eye precede and may cause the retinal degeneration. Indeed tim-fbl flies have diminished response to light transitions, and their altered day/night patterns of activity demonstrates defects in light perception. This suggest that retinal lesions are not solely due to oxidative stress and demonstrates a role for the transcriptional response to CoA deficiency underlying the defects observed in dPanK deficient flies. Moreover, in the present study we developed a new fly model that can be applied to other diseases and that allows the assessment of neurodegeneration in the brains of living flies.

Autism spectrum disorders (ASD) are neurodevelopmental disorders with phenotypic and genetic heterogeneity. Recent studies have reported rare and de novo mutations in ASD, but the allelic architecture of ASD remains unclear. To assess the role of common and rare variations in ASD, we constructed a gene co-expression network based on a widespread survey of gene expression in the human brain. We identified modules associated with specific cell types and processes. By integrating known rare mutations and the results of an ASD genome-wide association study (GWAS), we identified two neuronal modules that are perturbed by both rare and common variations. These modules contain highly connected genes that are involved in synaptic and neuronal plasticity and that are expressed in areas associated with learning and memory and sensory perception. The enrichment of common risk variants was replicated in two additional samples which include both simplex and multiplex families. An analysis of the combined contribution of common variants in the neuronal modules revealed a polygenic component to the risk of ASD. The results of this study point toward contribution of minor and major perturbations in the two sub-networks of neuronal genes to ASD risk.

Recombination events are not uniformly distributed and often cluster in narrow regions known as recombination hotspots. Several studies using different approaches have dramatically advanced our understanding of recombination hotspot regulation. Population genetic data have been used to map and quantify hotspots in the human genome. Genetic variation in recombination rates and hotspots usage have been explored in human pedigrees, mouse intercrosses, and by sperm typing. These studies pointed to the central role of the PRDM9 gene in hotspot modulation. In this study, we used single nucleotide polymorphisms (SNPs) from whole-genome resequencing and genotyping studies of mouse inbred strains to estimate recombination rates across the mouse genome and identified 47,068 historical hotspots--an average of over 2477 per chromosome. We show by simulation that inbred mouse strains can be used to identify positions of historical hotspots. Recombination hotspots were found to be enriched for the predicted binding sequences for different alleles of the PRDM9 protein. Recombination rates were on average lower near transcription start sites (TSS). Comparing the inferred historical recombination hotspots with the recent genome-wide mapping of double-strand breaks (DSBs) in mouse sperm revealed a significant overlap, especially toward the telomeres. Our results suggest that inbred strains can be used to characterize and study the dynamics of historical recombination hotspots. They also strengthen previous findings on mouse recombination hotspots, and specifically the impact of sequence variants in Prdm9.

Several different interventions improve depressed mood, including medication and environmental factors such as regular physical exercise. The molecular pathways underlying these effects are still not fully understood. In this study, we sought to identify shared mechanisms underlying antidepressant interventions. We studied three groups of mice: mice treated with a widely used antidepressant drug--fluoxetine, mice engaged in voluntary exercise, and mice living in an enriched environment. The hippocampi of treated mice were investigated at the molecular and cellular levels. Mice treated with fluoxetine and mice who exercised daily showed, not only similar antidepressant behavior, but also similar changes in gene expression and hippocampal neurons. These changes were not observed in mice with environmental enrichment. An increase in neurogenesis and dendritic spine density was observed following four weeks of fluoxetine treatment and voluntary exercise. A weighted gene co-expression network analysis revealed four different modules of co-expressed genes that were correlated with the antidepressant effect. This network analysis enabled us to identify genes involved in the molecular pathways underlying the effects of fluoxetine and exercise. The existence of both neuronal and gene expression changes common to antidepressant drug and exercise suggests a shared mechanism underlying their effect. Further studies of these findings may be used to uncover the molecular mechanisms of depression, and to identify new avenues of therapy.

Reelin plays an important role in the development and function of the brain and has been linked to different neuropsychiatric diseases. To further clarify the connection between reelin and psychiatric disorders, we studied the factors that influence the expression of reelin gene (RELN) and its different isoforms. We examined the total expression of RELN, allelic expression, and two alternative RELN isoforms in postmortem brain samples from patients with schizophrenia and bipolar disorder, as well as unaffected controls. We did not find a significant reduction in the total expression of RELN in schizophrenia or bipolar disorder. However, we did find a significant reduction of the proportion of the short RELN isoform, missing the C-terminal region in bipolar disorder, and imbalance in the allelic expression of RELN in schizophrenia. In addition, we tested the association between variation in RELN expression and rs7341475, an intronic SNP that was found to be associated with schizophrenia in women. We did not find an association between rs7341474 and the total expression of RELN either in women or in the entire sample. However, we observed a nominally significant effect of genotype-by-sex interaction on the variation in microexon skipping. Women with the risk genotype of rs7341475 (GG) had a higher proportion of microexon skipping, which is the isoform predominant in tissues outside the brain, while men had the opposite trend. Finally, we tested 83 SNPs in the gene region for association with expression variation of RELN, but none were significant. Our study further supports the connection between RELN dysfunction and psychiatric disorders, and provides a possible functional role for a schizophrenia associated SNP. Nevertheless, the positive associations observed in this study needs further replication as it may have implications for understanding the biological causes of schizophrenia and bipolar disorder.

Recent work has led to the identification of several susceptibility genes for autism spectrum disorder (ASD) and an increased appreciation of the importance of rare and de novo mutations. Some of the mutations may be very hard to detect using current strategies, especially if they are located in regulatory regions. We present a new approach to identify functional mutations that exploit the fact that many rare mutations disrupt the expression of genes from a single parental chromosome. The method incorporates measurement of the relative expression of the two copies of a gene across the genome using single nucleotide polymorphism arrays. Allelic expression has been successfully used to study common regulatory polymorphisms; however, it has not been implemented as a screening tool for rare mutation. We tested the potential of this approach by screening for monoallelic expression in lymphoblastoid cell lines derived from a small ASD cohort. After filtering regions shared across multiple samples, we identified genes showing monoallelic expression in specific ASD samples. Validation by quantitative sequencing demonstrated that the genes (or only part of them) are monoallelic expressed. The genes included both previously suspected risk factors for ASD and novel candidates. In one gene, named autism susceptibility candidate 2 (AUTS2), we identified a rare duplication that is likely to be the cause of monoallelic expression. Our results demonstrate the ability to identify rare regulatory mutations using genome-wide allelic expression screens, capabilities that could be expanded to other diseases, especially those with suspected involvement of rare dominantly acting mutations.

Genome-wide association studies (GWAS) with small sample size have had limited statistical power in identifying schizophrenia susceptibility genes. This is exemplified by the fact that one of the most convincing associations was detected only after meta-analyses of three different GWAS. Here we used meta-analysis to study the association of two single-nucleotide polymorphisms (SNPs) (rs7341475 and rs17746501) previously indicated to be associated with schizophrenia by a GWAS of Ashkenazi Jews (AJ). In the initial report, rs7341475 was associated only in women, while rs17746501 was associated in both men and women. We collected genotyping results of samples published in four GWAS for the two SNPs, additional to results from AJ. We used the Mantel-Haenszel method to combine the data of the different samples. For both SNPs, the results of the meta-analysis of all samples, including the initial report, did not reach a genome-wide significance level. However, the association between rs7341475 and schizophrenia in women, after excluding the data from AJ, was significant (P = 9.0 x 10(-3)), with a calculated odds ratio (OR) of 1.11, much smaller than the original result. Association between rs17746501 and schizophrenia was significant in four of the new samples, showing evidence for heterogeneity and very small effect when tested across all samples (P = 0.016, OR = 1.06). These findings suggest that the two SNPs might have a small effect on schizophrenia risk and suggest that meta-analyses of very large samples are needed to adequately study the contribution of common variants to schizophrenia susceptibility.