Abstract:BackgroundIn the United States, about 3 million people have autism spectrum disorder (ASD), and around 1 out of 59 children are diagnosed with ASD. People with ASD have characteristic social communication deficits and repetitive behaviors. The causes of this disorder remain unknown; however, in up to 25% of cases, a genetic cause can be identified. Detecting ASD as early as possible is desirable because early detection of ASD enables timely interventions in children with ASD. Identification of ASD based on objective pathogenic mutation screening is the major first step toward early intervention and effective treatment of affected children. ObjectiveRecent investigation interrogated genomics data for detecting and treating autism disorders, in addition to the conventional clinical interview as a diagnostic test. Since deep neural networks perform better than shallow machine learning models on complex and high-dimensional data, in this study, we sought to apply deep learning to genetic data obtained across thousands of simplex families at risk for ASD to identify contributory mutations and to create an advanced diagnostic classifier for autism screening. MethodsAfter preprocessing the genomics data from the Simons Simplex Collection, we extracted top ranking common variants that may be protective or pathogenic for autism based on a chi-square test. A convolutional neural network–based diagnostic classifier was then designed using the identified significant common variants to predict autism. The performance was then compared with shallow machine learning–based classifiers and randomly selected common variants. ResultsThe selected contributory common variants were significantly enriched in chromosome X while chromosome Y was also discriminatory in determining the identification of autistic individuals from nonautistic individuals. The ARSD, MAGEB16, and MXRA5 genes had the largest effect in the contributory variants. Thus, screening algorithms were adapted to include these common variants. The deep learning model yielded an area under the receiver operating characteristic curve of 0.955 and an accuracy of 88% for identifying autistic individuals from nonautistic individuals. Our classifier demonstrated a considerable improvement of ~13% in terms of classification accuracy compared to standard autism screening tools. ConclusionsCommon variants are informative for autism identification. Our findings also suggest that the deep learning process is a reliable method for distinguishing the diseased group from the control group based on the common variants of autism.

Placental DNA methylation changes and the early prediction of autism in full-term newborns

Diagnostic Classification and Prognostic Prediction Using Common Genetic Variants in Autism Spectrum Disorder: Genotype-Based Deep Learning

Sperm DNA methylation epimutation biomarker for paternal offspring autism susceptibility

Sex-specific DNA methylation signatures of autism spectrum disorder in newborn blood

Untargeted Metabolomics Screen of Mid-pregnancy Maternal Serum and Autism in Offspring

Evaluating the association between placenta DNA methylation and cognitive functions in the offspring

Epigenetic Regulation of Infant Neurobehavioral Outcomes

Artificial intelligence and bioinformatics analyze markers of children's transcriptional genome to predict autism spectrum disorder

Potential Role of Genomic Imprinted Genes and Brain Developmental Related Genes in Autism

Genome-wide DNA methylation profiles of autism spectrum disorder

Dysregulation of immune and metabolism pathways in maternal immune activation induces an increased risk of autism spectrum disorders

Prenatal Exposure of Diabetes and Progestin‐Mediated Autistic Biomarker in Peripheral Blood Mononuclear Cells

Epigenome-Wide Associations of Placental DNA Methylation and Behavioral and Emotional Difficulties in Children at 3 Years of Age

Molecular and Cellular Insights: A Focus on Glycans and the HNK1 Epitope in Autism Spectrum Disorder

Gene Regulation Analysis Reveals Perturbations of Autism Spectrum Disorder during Neural System Development

Prenatal Programming of Monocyte Chemotactic Protein-1 Signaling in Autism Susceptibility

Neurodevelopment and Metabolism in the Maternal-Placental-Fetal Unit

The intersection of inflammation and DNA damage as a novel axis underlying the pathogenesis of autism spectrum disorders

DNA Methylome Profiling of Maternal Peripheral Blood and Placentas Reveal Potential Fetal DNA Markers for Non-Invasive Prenatal Testing.

Epigenetic dysregulation of SHANK3 in brain tissues from individuals with autism spectrum disorders.

Integration of structural MRI and epigenetic analyses hint at linked cellular defects of the subventricular zone and insular cortex in autism: Findings from a case study