Application of deep learning algorithm on whole genome sequencing data uncovers structural variants associated with multiple mental disorders in African American patients

Abstract

Mental disorders present a global health concern, while the diagnosis of mental disorders can be challenging. The diagnosis is even harder for patients who have more than one type of mental disorder, especially for young toddlers who are not able to complete questionnaires or standardized rating scales for diagnosis. In the past decade, multiple genomic association signals have been reported for mental disorders, some of which present attractive drug targets. Concurrently, machine learning algorithms, especially deep learning algorithms, have been successful in the diagnosis and/or labeling of complex diseases, such as attention deficit hyperactivity disorder (ADHD) or cancer. In this study, we focused on eight common mental disorders, including ADHD, depression, anxiety, autism, intellectual disabilities, speech/language disorder, delays in developments, and oppositional defiant disorder in the ethnic minority of African Americans. Blood-derived whole genome sequencing data from 4179 individuals were generated, including 1384 patients with the diagnosis of at least one mental disorder. The burden of genomic variants in coding/non-coding regions was applied as feature vectors in the deep learning algorithm. Our model showed ~65% accuracy in differentiating patients from controls. Ability to label patients with multiple disorders was similarly successful, with a hamming loss score less than 0.3, while exact diagnostic matches are around 10%. Genes in genomic regions with the highest weights showed enrichment of biological pathways involved in immune responses, antigen/nucleic acid binding, chemokine signaling pathway, and G-protein receptor activities. A noticeable fact is that variants in non-coding regions (e.g., ncRNA, intronic, and intergenic) performed equally well as variants in coding regions; however, unlike coding region variants, variants in non-coding regions do not express genomic hotspots whereas they carry much more narrow standard deviations, indicating they probably serve as alternative markers.

[Retracted] Analysis of the Relevance Environment between Marxist Philosophy and System Theory Based on Deep Learning

“…The highest accuracy rate can reach 93.67%. Ref [ 19 ] model reduces the complexity of the model by reducing the number of network layers and the dimension of the feature representation layer, but at the same time, the accuracy rate is also reduced, but it is still better than the convolution network model. CNN model measures the time it takes for a pair of images to be recognized, which is also the standard to intuitively measure the complexity of the model.…”

“…Liu et al think that machine consciousness can be classified into five categories, and the operating mechanism of consciousness is explained in a frame. Material movement, physiological movement, neural activity, conscious activity, and intentional activity are an interrelated whole, pointing out that not all neural activities are conscious, and not all consciousness can clearly show intentionality [ 19 ]. Liu et al think that artificial intelligence realizes the revolution of thinking and poetry with the revolution of technology and method, and the top of intelligence opens the second paradigm of human reflection [ 20 ].…”

Genome Guided Personalized Drug Therapy in Attention Deficit Hyperactivity Disorder

“…The CNV burden of ADHD cases is similar to that found in autism spectrum disorders (ASD) or schizophrenia (19,20). Furthermore, recent exome wide genotyping (21) and sequencing and whole genome sequencing (WES/WGS), has shown many ADHD-related rare coding variants with much larger effects (22)(23)(24)(25)(26). Many of these variants and genes are established risk genes for other neuropsychiatric disorders.…”

DBFE: Distribution-based feature extraction from copy number and structural variants in whole-genome data

“…Deep learning approaches require large amounts of training samples, which is why they are most commonly applied only on readily available types of data, such as whole-exome sequencing, NGS panels, RNA-seq, or miRNA-seq. Such approaches are much rarer in whole-genome sequencing data analyses, where they are mainly used for population-wide studies (Liu et al ., 2022). When clinical information is required, e.g., in drug response prediction, deep learning approaches are likely to be inadequate because such information is scarce.…”