Summary: A new computational method has linked multiple target genes to autism spectrum disorder (ASD).
Source: University of Missouri, Columbia.
Autism spectrum disorder (ASD) encompasses a range of behavioral, social, and cognitive challenges. Early and accurate detection of ASD in children improves outcomes, but identifying the underlying genetic contributors is difficult because the disorder is highly heterogeneous. A multidisciplinary research team at the University of Missouri has developed a new computational approach that connects specific genes and gene combinations to subgroups of autism. These findings could inform future screening tools for young children and help clinicians choose more appropriate, tailored interventions.
Tracing genetic causes in autism requires extensive data analysis. In 2014 the National Science Foundation awarded a $1 million grant to MU to install high-performance computing resources that support large-scale bioinformatics and data-driven engineering research and education. Those computing capabilities made possible the intensive analyses performed in this study.
“We began with more than 2,591 families in which only one child had been diagnosed with autism while neither the parents nor siblings showed a diagnosis,” said Chi-Ren Shyu, director of the Informatics Institute and the Paul K. and Dianne Shumaker Endowed Professor in the MU Department of Electrical Engineering and Computer Science. “That created a genetically diverse dataset with roughly ten million variants. We reduced that set to about 30,000 of the most promising variants, then applied tailored algorithms and the university’s high-performance computing to mine those genetic signals.”
Genetic samples were provided by the Simons Foundation Autism Research Initiative. The dataset included more than 11,500 individuals—children diagnosed with autism and their unaffected parents and siblings. Using advanced computational techniques, Shyu and collaborators isolated 286 genes associated with defined subgroups of autism. Those genes were organized into 12 subgroups that reflected commonly observed trait patterns among individuals on the spectrum. Of the 286 genes, 193 had not appeared as autism candidates in prior studies.
“Autism is heterogeneous, which means the genetic contributors vary widely,” said Judith Miles, professor emerita of child health-genetics at the MU Thompson Center for Autism and Neurodevelopmental Disorders. “That variability makes it difficult for traditional genetic approaches to pinpoint causes. The informatics strategies developed by Dr. Shyu and the results our team found provide geneticists with many new candidate targets. By narrowing the set of relevant genetic markers, we can start developing clinical programs and diagnostic methods that are more precise. This is a significant step forward in understanding genetic factors in autism.”
The informatics framework demonstrated in this work is scalable and ready for larger datasets, for example the Simons Foundation’s SPARK initiative, which aims to collect genetic and clinical information from up to 50,000 individuals with autism and their families. That scale would enable deeper exploration of gene associations and faster progress toward understanding the causes of autism and identifying supports and treatments. SPARK is an ongoing national research effort that partners with families to build a research resource for scientists.
Contributors to the study included Matt Spencer from the MU Informatics Institute; Nicole Takahashi from the MU Thompson Center for Autism; and Sounak Chakraborty from the MU Department of Statistics.
Funding: This research was supported by the National Institutes of Health (5T32GM008396, 5T32LM012410-02); the Shumaker Endowment for Biomedical Informatics; the National Science Foundation (CNS-1429294); and the Simons Foundation.
Source: Jeff Sossamon, University of Missouri, Columbia.
Publisher: NeuroscienceNews.com.
Image credit: MU Thompson Center for Autism and Neurodevelopmental Disorders.
Original research: Published in the Journal of Biomedical Informatics. DOI: 10.1016/j.jbi.2017.11.016.
Heritable genotype contrast mining reveals novel gene associations specific to autism subgroups
Autism’s genetic architecture is complex, but progress can be made by identifying genes and gene-gene interactions that contribute to distinct autism subtypes. Recognizing these groupings can enable diagnosis and treatment tailored to specific trait profiles. The team developed a computational pipeline that prioritizes genetic variants across the genome and applies frequent pattern mining to detect potential interactions among variants. A novel genotype metric, the Unique Inherited Combination support, accounts for inheritance patterns within nuclear families and estimates how variant combinations influence phenotypes at the individual level. High-contrast variant combinations are then examined for significant associations with subgroups. By contrasting subgroups defined by severe or mild manifestations of particular traits, the method connected 286 genes to specific subgroups, including 193 candidate genes not previously linked to autism. The analysis also revealed 71 gene pairs with joint subgroup associations, suggesting interacting pathways worth investigating. Although this study focused on 12 autism subgroups, the method can be applied to other clinically meaningful subdivisions of ASD and to other phenotypically heterogeneous disorders, such as Alzheimer’s disease.
This article summarizes university-reported research findings and the computational approach used to prioritize and associate genetic variants with autism subgroups. The findings highlight new candidate genes and gene combinations that can guide future genetic studies, improve early screening strategies, and refine intervention approaches for children diagnosed with ASD.