30902
Comparative Analysis of ASD Gene Prioritization Strategies
Identification of a core set of highly penetrant and causative ASD genes among the growing list of genes associated with the disorder has become a central goal in autism research. To that end, we have developed a gene-scoring algorithm that relies on ongoing curation of individual ASD-associated variants in the autism genetic database (AutDB). The assessment of ASD-associated variants is based on multiple parameters such as significance of genetic association, family structure, inheritance pattern, zygosity, variant type, and functional effect (Larsen et al., 2016).
Objectives:
Our gene-scoring algorithm is one of several prioritization strategies that have emerged to aid in the identification of a core set of likely pathogenic ASD genes. However, there are differences between the approaches by which these strategies identify their own unique gene sets. Here, we utilized a modified version of our gene-scoring algorithm and compared our findings to other available ASD candidate gene datasets.
Methods:
We analyzed a total of 13,120 individual variants in 1036 genes associated with ASD that were annotated from 3544 research articles (AutDB data freeze of September 2018) in order to prioritize ASD candidate genes and identify a set of high-confidence genes. We then compared the results of our prioritization analysis to five other available ASD gene datasets: the SFARI Gene Scoring Module (data freeze of September 2018); a dataset of 65 ASD candidate genes identified by TADA analysis of whole-exome sequencing data (Sanders et al., 2015); candidate genes identified by a human brain-specific functional interaction network (Krishnan et al., 2016); the Developmental Brain Database (Gonzales-Mantilla et al., 2016); and the 78 genes comprising the SPARK gene list (updated April 2017).
Results:
The majority of the highest ranking ASD genes identified by our gene-scoring algorithm were well represented in four out of five datasets analyzed in this study. However, we also observed significant differences between our results and the other datasets, in part due to the types of genetic evidence used in assessing the importance of a gene in ASD pathogenesis, as well as to the dynamic assessment of candidate genes made possible by real-time curation of ASD-associated variants in AutDB. Finally, we observed little overlap between our results and those reported in Krishnan et al., 2016, with only 5 of our high confidence ASD genes ranked within the top 100 genes identified by interaction networks.
Conclusions:
The gene prioritization strategies used here help to confirm the importance of a set of genes well-represented by other authors and other methodologies while also highlighting genes for which some strong evidence exists that is suggestive of their causative role in autism but which have been overlooked by other methods.