A Machine Learning Approach to Predicting Autism Risk Genes: Validation of Known Genes and Discovery of New Candidates

ASiDentify (ASiD): a machine learning model to predict new autism spectrum disorder risk genes

Autism spectrum disorder (ASD) is a neurodevelopmental disorder that affects nearly 3% of children and has a strong genetic component. While hundreds of ASD risk genes have been identified through sequencing studies, the genetic heterogeneity of ASD makes identifying additional risk genes using these methods challenging. To predict candidate ASD risk genes, we developed a simple machine learning model, ASiDentify (ASiD), using human genomic, RNA- and protein-based features. ASiD identified over 1,300 candidate ASD risk genes, over 300 of which have not been previously predicted. ASiD made accurate predictions of ASD risk genes using 6 features predictive of ASD risk gene status, including mutational constraint, synapse localization and gene expression in neurons, astrocytes and non-brain tissues. Particular functional groups of proteins found to be strongly implicated in ASD include RNA-binding proteins (RBPs) and chromatin regulators. We constructed additional logistic regression models to make predictions and assess informative features specific to RBPs, including mutational constraint, or chromatin regulators, for which both expression level in excitatory neurons and mutational constraint were informative. The fact that RBPs and chromatin regulators had informative features distinct from all protein-coding genes suggests that specific biological pathways connect risk genes with different molecular functions to ASD.

Germline mosaicism in TCF20-associated neurodevelopmental disorders: a case study and literature review

Autosomal dominant variants in transcription factor 20 (TCF20) can result in TCF20-associated neurodevelopmental disorder (TAND), a condition characterized by developmental delay and intellectual disability, autism, dysmorphisms, dystonia, and variable other neurological features. To date, a total of 91 individuals with TAND have been reported; ~67% of cases arose de novo, while ~10% were inherited, and, intriguingly, ~8% were either confirmed or suspected to have arisen via germline mosaicism. Here, we describe two siblings with a developmental condition characterized by intellectual disability, autism, a circadian rhythm sleep disorder, and attention deficit hyperactivity disorder (ADHD) caused by a novel heterozygous single nucleotide deletion in the TCF20 gene, NM_001378418.1:c.4737del; NP_001365347.1:p.Lys1579Asnfs*36 (GRCh38/hg38). The variant was not detected in DNA extracted from peripheral blood in either parent by Sanger sequencing of PCR-generated amplicons, or by deep sequencing of PCR amplicons using MiSeq and MinION. However, droplet digital PCR (ddPCR) of DNA derived from early morning urine detected the variation in 3.2% of the father's urothelial cells, confirming germline mosaicism. This report is only the second to confirm with physical evidence TCF20 germline mosaicism and discusses germline mosaicism as a likely under-detected mode of inheritance in neurodevelopmental conditions.

Sequence variants in HECTD1 result in a variable neurodevelopmental disorder

Dysregulation of genes encoding the homologous to E6AP C-terminus (HECT) E3 ubiquitin ligases has been linked to cancer and structural birth defects. One member of this family, the HECT-domain-containing protein 1 (HECTD1), mediates developmental pathways, including cell signaling, gene expression, and embryogenesis. Through GeneMatcher, we identified 14 unrelated individuals with 15 different variants in HECTD1 (10 missense, 3 frameshift, 1 nonsense, and 1 splicing variant) with neurodevelopmental disorders (NDDs), including autism, attention-deficit/hyperactivity disorder, and epilepsy. Of these 15 HECTD1 variants, 10 occurred de novo, 3 had unknown inheritance, and 2 were compound heterozygous. While all individuals in this cohort displayed NDDs, no genotype-phenotype correlation was apparent. Conditional knockout of Hectd1 in the neural lineage in mice resulted in microcephaly, severe hippocampal malformations, and complete agenesis of the corpus callosum, supporting a role for Hectd1 in embryonic brain development. Functional studies of select variants in C. elegans revealed dominant effects, including either change-of-function or loss-of-function/haploinsufficient mechanisms, which may explain phenotypic heterogeneity. Significant enrichment of de novo variants in HECTD1 was also shown in an independent cohort of 53,305 published trios with NDDs or congenital heart disease. Thus, our clinical and functional data support a critical requirement of HECTD1 for human brain development.

Proximity analysis of native proteomes reveals phenotypic modifiers in a mouse model of autism and related neurodevelopmental conditions

One of the main drivers of autism spectrum disorder is risk alleles within hundreds of genes, which may interact within shared but unknown protein complexes. Here we develop a scalable genome-editing-mediated approach to target 14 high-confidence autism risk genes within the mouse brain for proximity-based endogenous proteomics, achieving the identification of high-specificity spatial proteomes. The resulting native proximity proteomes are enriched for human genes dysregulated in the brain of autistic individuals, and reveal proximity interactions between proteins from high-confidence risk genes with those of lower-confidence that may provide new avenues to prioritize genetic risk. Importantly, the datasets are enriched for shared cellular functions and genetic interactions that may underlie the condition. We test this notion by spatial proteomics and CRISPR-based regulation of expression in two autism models, demonstrating functional interactions that modulate mechanisms of their dysregulation. Together, these results reveal native proteome networks in vivo relevant to autism, providing new inroads for understanding and manipulating the cellular drivers underpinning its etiology.

The Importance of Large-Scale Genomic Studies to Unravel Genetic Risk Factors for Autism

Autism spectrum disorder (ASD) is a common and highly heritable neurodevelopmental disorder. During the last 15 years, advances in genomic technologies and the availability of increasingly large patient cohorts have greatly expanded our knowledge of the genetic architecture of ASD and its neurobiological mechanisms. Over two hundred risk regions and genes carrying rare de novo and transmitted high-impact variants have been identified. Additionally, common variants with small individual effect size are also important, and a number of loci are now being uncovered. At the same time, these new insights have highlighted ongoing challenges. In this perspective article, we summarize developments in ASD genetic research and address the enormous impact of large-scale genomic initiatives on ASD gene discovery.

A network-based method for associating genes with autism spectrum disorder

Autism spectrum disorder (ASD) is a highly heritable complex disease that affects 1% of the population, yet its underlying molecular mechanisms are largely unknown. Here we study the problem of predicting causal genes for ASD by combining genome-scale data with a network propagation approach. We construct a predictor that integrates multiple omic data sets that assess genomic, transcriptomic, proteomic, and phosphoproteomic associations with ASD. In cross validation our predictor yields mean area under the ROC curve of 0.87 and area under the precision-recall curve of 0.89. We further show that it outperforms previous gene-level predictors of autism association. Finally, we show that we can use the model to predict genes associated with Schizophrenia which is known to share genetic components with ASD.

Systematic bibliometric and visualized analysis of research hotspots and trends in artificial intelligence in autism spectrum disorder

Background

Artificial intelligence (AI) has been the subject of studies in autism spectrum disorder (ASD) and may affect its identification, diagnosis, intervention, and other medical practices in the future. Although previous studies have used bibliometric techniques to analyze and investigate AI, there has been little research on the adoption of AI in ASD. This study aimed to explore the broad applications and research frontiers of AI used in ASD.

Methods

Citation data were retrieved from the Web of Science Core Collection (WoSCC) database to assess the extent to which AI is used in ASD. CiteSpace.5.8. R3 and VOSviewer, two online tools for literature metrology analysis, were used to analyze the data.

Results

A total of 776 publications from 291 countries and regions were analyzed; of these, 256 publications were from the United States and 173 publications were from China, and England had the largest centrality of 0.33; Stanford University had the highest H-index of 17; and the largest cluster label of co-cited references was machine learning. In addition, keywords with a high number of occurrences in this field were autism spectrum disorder (295), children (255), classification (156) and diagnosis (77). The burst keywords from 2021 to 2023 were infants and feature selection, and from 2022 to 2023, the burst keyword was corpus callosum.

Conclusion

This research provides a systematic analysis of the literature concerning AI used in ASD, presenting an overall demonstration in this field. In this area, the United States and China have the largest number of publications, England has the greatest influence, and Stanford University is the most influential. In addition, the research on AI used in ASD mostly focuses on classification and diagnosis, and "infants, feature selection, and corpus callosum are at the forefront, providing directions for future research. However, the use of AI technologies to identify ASD will require further research.

Neuroimaging genetics approaches to identify new biomarkers for the early diagnosis of autism spectrum disorder

Autism-spectrum disorders (ASDs) are developmental disabilities that manifest in early childhood and are characterized by qualitative abnormalities in social behaviors, communication skills, and restrictive or repetitive behaviors. To explore the neurobiological mechanisms in ASD, extensive research has been done to identify potential diagnostic biomarkers through a neuroimaging genetics approach. Neuroimaging genetics helps to identify ASD-risk genes that contribute to structural and functional variations in brain circuitry and validate biological changes by elucidating the mechanisms and pathways that confer genetic risk. Integrating artificial intelligence models with neuroimaging data lays the groundwork for accurate diagnosis and facilitates the identification of early diagnostic biomarkers for ASD. This review discusses the significance of neuroimaging genetics approaches to gaining a better understanding of the perturbed neurochemical system and molecular pathways in ASD and how these approaches can detect structural, functional, and metabolic changes and lead to the discovery of novel biomarkers for the early diagnosis of ASD.

Dysfunction of ubiquitin protein ligase MYCBP2 leads to cell resilience in human breast cancers

Breast cancer is the most common type of cancer among women worldwide, and it is estimated that 294 000 new diagnoses and 37 000 deaths will occur each year in the United States alone by 2030. Large-scale genomic studies have identified a number of genetic loci with alterations in breast cancer. However, identification of the genes that are critical for tumorgenicity still remains a challenge. Here, we perform a comprehensive functional multi-omics analysis of somatic mutations in breast cancer and identify previously unknown key regulators of breast cancer tumorgenicity. We identify dysregulation of MYCBP2, an E3 ubiquitin ligase and an upstream regulator of mTOR signaling, is accompanied with decreased disease-free survival. We validate MYCBP2 as a key target through depletion siRNA using in vitro apoptosis assays in MCF10A, MCF7 and T47D cells. We demonstrate that MYCBP2 loss is associated with resistance to apoptosis from cisplatin-induced DNA damage and cell cycle changes, and that CHEK1 inhibition can modulate MYCBP2 activity and caspase cleavage. Furthermore, we show that MYCBP2 knockdown is associated with transcriptomic responses in TSC2 and in apoptosis genes and interleukins. Therefore, we show that MYCBP2 is an important genetic target that represents a key node regulating multiple molecular pathways in breast cancer corresponding with apparent drug resistance in our study.

Tcf20 deficiency is associated with increased liver fibrogenesis and alterations in mitochondrial metabolism in mice and humans

BACKGROUND & AIMS

Transcription co-activator factor 20 (TCF20) is a regulator of transcription factors involved in extracellular matrix remodelling. In addition, TCF20 genomic variants in humans have been associated with impaired intellectual disability. Therefore, we hypothesized that TCF20 has several functions beyond those described in neurogenesis, including the regulation of fibrogenesis.

METHODS

Tcf20 knock-out (Tcf20-/- ) and Tcf20 heterozygous mice were generated by homologous recombination. TCF20 gene genotyping and expression was assessed in patients with pathogenic variants in the TCF20 gene. Neural development was investigated by immufluorescense. Mitochondrial metabolic activity was evaluated with the Seahorse analyser. The proteome analysis was carried out by gas chromatography mass-spectrometry.

RESULTS

Characterization of Tcf20-/- newborn mice showed impaired neural development and death after birth. In contrast, heterozygous mice were viable but showed higher CCl4 -induced liver fibrosis and a differential expression of genes involved in extracellular matrix homeostasis compared to wild-type mice, along with abnormal behavioural patterns compatible with autism-like phenotypes. Tcf20-/- embryonic livers and mouse embryonic fibroblast (MEF) cells revealed differential expression of structural proteins involved in the mitochondrial oxidative phosphorylation chain, increased rates of mitochondrial metabolic activity and alterations in metabolites of the citric acid cycle. These results parallel to those found in patients with TCF20 pathogenic variants, including alterations of the fibrosis scores (ELF and APRI) and the elevation of succinate concentration in plasma.

CONCLUSIONS

We demonstrated a new role of Tcf20 in fibrogenesis and mitochondria metabolism in mice and showed the association of TCF20 deficiency with fibrosis and metabolic biomarkers in humans.

The Lost and Found: Unraveling the Functions of Orphan Genes

Orphan Genes (OGs) are a mysterious class of genes that have recently gained significant attention. Despite lacking a clear evolutionary history, they are found in nearly all living organisms, from bacteria to humans, and they play important roles in diverse biological processes. The discovery of OGs was first made through comparative genomics followed by the identification of unique genes across different species. OGs tend to be more prevalent in species with larger genomes, such as plants and animals, and their evolutionary origins remain unclear but potentially arise from gene duplication, horizontal gene transfer (HGT), or de novo origination. Although their precise function is not well understood, OGs have been implicated in crucial biological processes such as development, metabolism, and stress responses. To better understand their significance, researchers are using a variety of approaches, including transcriptomics, functional genomics, and molecular biology. This review offers a comprehensive overview of the current knowledge of OGs in all domains of life, highlighting the possible role of dark transcriptomics in their evolution. More research is needed to fully comprehend the role of OGs in biology and their impact on various biological processes.

A self-inspected adaptive SMOTE algorithm (SASMOTE) for highly imbalanced data classification in healthcare

In many healthcare applications, datasets for classification may be highly imbalanced due to the rare occurrence of target events such as disease onset. The SMOTE (Synthetic Minority Over-sampling Technique) algorithm has been developed as an effective resampling method for imbalanced data classification by oversampling samples from the minority class. However, samples generated by SMOTE may be ambiguous, low-quality and non-separable with the majority class. To enhance the quality of generated samples, we proposed a novel self-inspected adaptive SMOTE (SASMOTE) model that leverages an adaptive nearest neighborhood selection algorithm to identify the "visible" nearest neighbors, which are used to generate samples likely to fall into the minority class. To further enhance the quality of the generated samples, an uncertainty elimination via self-inspection approach is introduced in the proposed SASMOTE model. Its objective is to filter out the generated samples that are highly uncertain and inseparable with the majority class. The effectiveness of the proposed algorithm is compared with existing SMOTE-based algorithms and demonstrated through two real-world case studies in healthcare, including risk gene discovery and fatal congenital heart disease prediction. By generating the higher quality synthetic samples, the proposed algorithm is able to help achieve better prediction performance (in terms of F1 score) on average compared to the other methods, which is promising to enhance the usability of machine learning models on highly imbalanced healthcare data.

Regulation of MAPK Signaling Pathways by the Large HERC Ubiquitin Ligases

Protein ubiquitylation acts as a complex cell signaling mechanism since the formation of different mono- and polyubiquitin chains determines the substrate's fate in the cell. E3 ligases define the specificity of this reaction by catalyzing the attachment of ubiquitin to the substrate protein. Thus, they represent an important regulatory component of this process. Large HERC ubiquitin ligases belong to the HECT E3 protein family and comprise HERC1 and HERC2 proteins. The physiological relevance of the Large HERCs is illustrated by their involvement in different pathologies, with a notable implication in cancer and neurological diseases. Understanding how cell signaling is altered in these different pathologies is important for uncovering novel therapeutic targets. To this end, this review summarizes the recent advances in how the Large HERCs regulate the MAPK signaling pathways. In addition, we emphasize the potential therapeutic strategies that could be followed to ameliorate the alterations in MAPK signaling caused by Large HERC deficiencies, focusing on the use of specific inhibitors and proteolysis-targeting chimeras.

Machine Learning-Based Blood RNA Signature for Diagnosis of Autism Spectrum Disorder

Early diagnosis of autism spectrum disorder (ASD) is crucial for providing appropriate treatments and parental guidance from an early age. Yet, ASD diagnosis is a lengthy process, in part due to the lack of reliable biomarkers. We recently applied RNA-sequencing of peripheral blood samples from 73 American and Israeli children with ASD and 26 neurotypically developing (NT) children to identify 10 genes with dysregulated blood expression levels in children with ASD. Machine learning (ML) analyzes data by computerized analytical model building and may be applied to building diagnostic tools based on the optimization of large datasets. Here, we present several ML-generated models, based on RNA expression datasets collected during our recently published RNA-seq study, as tentative tools for ASD diagnosis. Using the random forest classifier, two of our proposed models yield an accuracy of 82% in distinguishing children with ASD and NT children. Our proof-of-concept study requires refinement and independent validation by studies with far larger cohorts of children with ASD and NT children and should thus be perceived as starting point for building more accurate ML-based tools. Eventually, such tools may potentially provide an unbiased means to support the early diagnosis of ASD.

HERC1 deficiency causes osteopenia through transcriptional program dysregulation during bone remodeling

Bone remodeling is a continuous process between bone-forming osteoblasts and bone-resorbing osteoclasts, with any imbalance resulting in metabolic bone disease, including osteopenia. The HERC1 gene encodes an E3 ubiquitin ligase that affects cellular processes by regulating the ubiquitination of target proteins, such as C-RAF. Of interest, an association exists between biallelic pathogenic sequence variants in the HERC1 gene and the neurodevelopmental disorder MDFPMR syndrome (macrocephaly, dysmorphic facies, and psychomotor retardation). Most pathogenic variants cause loss of HERC1 function, and the affected individuals present with features related to altered bone homeostasis. Herc1-knockout mice offer an excellent model in which to study the role of HERC1 in bone remodeling and to understand its role in disease. In this study, we show that HERC1 regulates osteoblastogenesis and osteoclastogenesis, proving that its depletion increases gene expression of osteoblastic makers during the osteogenic differentiation of mesenchymal stem cells. During this process, HERC1 deficiency increases the levels of C-RAF and of phosphorylated ERK and p38. The Herc1-knockout adult mice developed imbalanced bone homeostasis that presented as osteopenia in both sexes of the adult mice. By contrast, only young female knockout mice had osteopenia and increased number of osteoclasts, with the changes associated with reductions in testosterone and dihydrotestosterone levels. Finally, osteocytes isolated from knockout mice showed a higher expression of osteocytic genes and an increase in the Rankl/Opg ratio, indicating a relevant cell-autonomous role of HERC1 when regulating the transcriptional program of bone formation. Overall, these findings present HERC1 as a modulator of bone homeostasis and highlight potential therapeutic targets for individuals affected by pathological HERC1 variants.

The HERC proteins and the nervous system

The HERC protein family is one of three subfamilies of Homologous to E6AP C-terminus (HECT) E3 ubiquitin ligases. Six HERC genes have been described in humans, two of which encode Large HERC proteins -HERC1 and HERC2- with molecular weights above 520 kDa that are constitutively expressed in the brain. There is a large body of evidence that mutations in these Large HERC genes produce clinical syndromes in which key neurodevelopmental events are altered, resulting in intellectual disability and other neurological disorders like epileptic seizures, dementia and/or signs of autism. In line with these consequences in humans, two mice carrying mutations in the Large HERC genes have been studied quite intensely: the tambaleante mutant for Herc1 and the Herc2^+/530 mutant for Herc2. In both these mutant mice there are clear signs that autophagy is dysregulated, eliciting cerebellar Purkinje cell death and impairing motor control. The tambaleante mouse was the first of these mice to appear and is the best studied, in which the Herc1 mutation elicits: (i) delayed neural transmission in the peripheral nervous system; (ii) impaired learning, memory and motor control; and (iii) altered presynaptic membrane dynamics. In this review, we discuss the information currently available on HERC proteins in the nervous system and their biological activity, the dysregulation of which could explain certain neurodevelopmental syndromes and/or neurodegenerative diseases.

De novo mutations within metabolism networks of amino acid/protein/energy in Chinese autistic children with intellectual disability

BACKGROUND

Autism spectrum disorder (ASD) is often accompanied by intellectual disability (ID). Despite extensive studies, however, the genetic basis for this comorbidity is still not clear. In this study, we tried to develop an analyzing pipeline for de novo mutations and possible pathways related to ID phenotype in ASD. Whole-exome sequencing (WES) was performed to screen de novo mutations and candidate genes in 79 ASD children together with their parents (trios). The de novo altering genes and relative pathways which were associated with ID phenotype were analyzed. The connection nodes (genes) of above pathways were selected, and the diagnostic value of these selected genes for ID phenotype in the study population was also evaluated.

RESULTS

We identified 89 de novo mutant genes, of which 34 genes were previously reported to be associated with ASD, including double hits in the EGF repeats of NOTCH1 gene (p.V999M and p.S1027L). Interestingly, of these 34 genes, 22 may directly affect intelligence quotient (IQ). Further analyses revealed that these IQ-related genes were enriched in protein synthesis, energy metabolism, and amino acid metabolism, and at least 9 genes (CACNA1A, ALG9, PALM2, MGAT4A, PCK2, PLEKHA1, PSME3, ADI1, and TLE3) were involved in all these three pathways. Seven patients who harbored these gene mutations showed a high prevalence of a low IQ score (< 70), a non-verbal language, and an early diagnostic age (< 4 years). Furthermore, our panel of these 9 genes reached a 10.2% diagnostic rate (5/49) in early diagnostic patients with a low IQ score and also reached a 10% diagnostic yield in those with both a low IQ score and non-verbal language (4/40).

CONCLUSION

We found some new genetic disposition for ASD accompanied with intellectual disability in this study. Our results may be helpful for etiologic research and early diagnoses of intellectual disability in ASD. Larger population studies and further mechanism studies are warranted.

Triage and priority-based healthcare diagnosis using artificial intelligence for autism spectrum disorder and gene contribution: A systematic review

The exact nature, harmful effects and aetiology of autism spectrum disorder (ASD) have caused widespread confusion. Artificial intelligence (AI) science helps solve challenging diagnostic problems in the medical field through extensive experiments. Disease severity is closely related to triage decisions and prioritisation contexts in medicine because both have been widely used to diagnose various diseases via AI, machine learning and automated decision-making techniques. Recently, taking advantage of high-performance AI algorithms has achieved accessible success in diagnosing and predicting risks from clinical and biological data. In contrast, less progress has been made with ASD because of obscure reasons. According to academic literature, ASD diagnosis works from a specific perspective, and much of the confusion arises from the fact that how AI techniques are currently integrated with the diagnosis of ASD concerning the triage and priority strategies and gene contributions. To this end, this study sought to describe a systematic review of the literature to assess the respective AI methods using the available datasets, highlight the tools and strategies used for diagnosing ASD and investigate how AI trends contribute in distinguishing triage and priority for ASD and gene contributions. Accordingly, this study checked the Science Direct, IEEE Xplore Digital Library, Web of Science (WoS), PubMed, and Scopus databases. A set of 363 articles from 2017 to 2022 is collected to reveal a clear picture and a better understanding of all the academic literature through a final set of 18 articles. The retrieved articles were filtered according to the defined inclusion and exclusion criteria and classified into three categories. The first category includes 'Triage patients based on diagnosis methods' which accounts for 16.66% (n = 3/18). The second category includes 'Prioritisation for Risky Genes' which accounts for 66.6% (n = 12/18) and is classified into two subcategories: 'Mutations observation based', 'Biomarkers and toxic chemical observations'. The third category includes 'E-triage using telehealth' which accounts for 16.66% (n = 3/18). This multidisciplinary systematic review revealed the taxonomy, motivations, recommendations and challenges of ASD research that need synergistic attention. Thus, this systematic review performs a comprehensive science mapping analysis and discusses the open issues that help perform and improve the recommended solution of ASD research direction. In addition, this study critically reviews the literature and attempts to address the current research gaps in knowledge and highlights weaknesses that require further research. Finally, a new developed methodology has been suggested as future work for triaging and prioritising ASD patients according to their severity levels by using decision-making techniques.

SFARI genes and where to find them; modelling Autism Spectrum Disorder specific gene expression dysregulation with RNA-seq data

Autism Spectrum Disorders (ASD) have a strong, yet heterogeneous, genetic component. Among the various methods that are being developed to help reveal the underlying molecular aetiology of the disease one approach that is gaining popularity is the combination of gene expression and clinical genetic data, often using the SFARI-gene database, which comprises lists of curated genes considered to have causative roles in ASD when mutated in patients. We build a gene co-expression network to study the relationship between ASD-specific transcriptomic data and SFARI genes and then analyse it at different levels of granularity. No significant evidence is found of association between SFARI genes and differential gene expression patterns when comparing ASD samples to a control group, nor statistical enrichment of SFARI genes in gene co-expression network modules that have a strong correlation with ASD diagnosis. However, classification models that incorporate topological information from the whole ASD-specific gene co-expression network can predict novel SFARI candidate genes that share features of existing SFARI genes and have support for roles in ASD in the literature. A statistically significant association is also found between the absolute level of gene expression and SFARI's genes and Scores, which can confound the analysis if uncorrected. We propose a novel approach to correct for this that is general enough to be applied to other problems affected by continuous sources of bias. It was found that only co-expression network analyses that integrate information from the whole network are able to reveal signatures linked to ASD diagnosis and novel candidate genes for the study of ASD, which individual gene or module analyses fail to do. It was also found that the influence of SFARI genes permeates not only other ASD scoring systems, but also lists of genes believed to be involved in other neurodevelopmental disorders.

Association of CDH11 with Autism Spectrum Disorder Revealed by Matched-gene Co-expression Analysis and Mouse Behavioral Studies

A large number of putative risk genes for autism spectrum disorder (ASD) have been reported. The functions of most of these susceptibility genes in developing brains remain unknown, and causal relationships between their variation and autism traits have not been established. The aim of this study was to predict putative risk genes at the whole-genome level based on the analysis of gene co-expression with a group of high-confidence ASD risk genes (hcASDs). The results showed that three gene features - gene size, mRNA abundance, and guanine-cytosine content - affect the genome-wide co-expression profiles of hcASDs. To circumvent the interference of these features in gene co-expression analysis, we developed a method to determine whether a gene is significantly co-expressed with hcASDs by statistically comparing the co-expression profile of this gene with hcASDs to that of this gene with permuted gene sets of feature-matched genes. This method is referred to as "matched-gene co-expression analysis" (MGCA). With MGCA, we demonstrated the convergence in developmental expression profiles of hcASDs and improved the efficacy of risk gene prediction. The results of analysis of two recently-reported ASD candidate genes, CDH11 and CDH9, suggested the involvement of CDH11, but not CDH9, in ASD. Consistent with this prediction, behavioral studies showed that Cdh11-null mice, but not Cdh9-null mice, have multiple autism-like behavioral alterations. This study highlights the power of MGCA in revealing ASD-associated genes and the potential role of CDH11 in ASD.

"Guilt by association" is not competitive with genetic association for identifying autism risk genes

Discovering genes involved in complex human genetic disorders is a major challenge. Many have suggested that machine learning (ML) algorithms using gene networks can be used to supplement traditional genetic association-based approaches to predict or prioritize disease genes. However, questions have been raised about the utility of ML methods for this type of task due to biases within the data, and poor real-world performance. Using autism spectrum disorder (ASD) as a test case, we sought to investigate the question: can machine learning aid in the discovery of disease genes? We collected 13 published ASD gene prioritization studies and evaluated their performance using known and novel high-confidence ASD genes. We also investigated their biases towards generic gene annotations, like number of association publications. We found that ML methods which do not incorporate genetics information have limited utility for prioritization of ASD risk genes. These studies perform at a comparable level to generic measures of likelihood for the involvement of genes in any condition, and do not out-perform genetic association studies. Future efforts to discover disease genes should be focused on developing and validating statistical models for genetic association, specifically for association between rare variants and disease, rather than developing complex machine learning methods using complex heterogeneous biological data with unknown reliability.

A Review of Machine Learning Methods of Feature Selection and Classification for Autism Spectrum Disorder

Autism Spectrum Disorder (ASD), according to DSM-5 in the American Psychiatric Association, is a neurodevelopmental disorder that includes deficits of social communication and social interaction with the presence of restricted and repetitive behaviors. Children with ASD have difficulties in joint attention and social reciprocity, using non-verbal and verbal behavior for communication. Due to these deficits, children with autism are often socially isolated. Researchers have emphasized the importance of early identification and early intervention to improve the level of functioning in language, communication, and well-being of children with autism. However, due to limited local assessment tools to diagnose these children, limited speech-language therapy services in rural areas, etc., these children do not get the rehabilitation they need until they get into compulsory schooling at the age of seven years old. Hence, efficient approaches towards early identification and intervention through speedy diagnostic procedures for ASD are required. In recent years, advanced technologies like machine learning have been used to analyze and investigate ASD to improve diagnostic accuracy, time, and quality without complexity. These machine learning methods include artificial neural networks, support vector machines, a priori algorithms, and decision trees, most of which have been applied to datasets connected with autism to construct predictive models. Meanwhile, the selection of features remains an essential task before developing a predictive model for ASD classification. This review mainly investigates and analyzes up-to-date studies on machine learning methods for feature selection and classification of ASD. We recommend methods to enhance machine learning's speedy execution for processing complex data for conceptualization and implementation in ASD diagnostic research. This study can significantly benefit future research in autism using a machine learning approach for feature selection, classification, and processing imbalanced data.