Author(s): Kyle A. Sullivan, Oak Ridge National Laboratory, Computational and Predictive Biology, Oak Ridge, TN; David Kainer, Oak Ridge National Laboratory, Computational and Predictive Biology, Oak Ridge, TN; Xuejun Qin, Duke University School of Medicine, Duke University, Durham, NC; Durham Veterans Affairs Medical Center, Durham, NC; Jennifer Lindquist, Duke University School of Medicine, Duke University, Durham, NC; Durham Veterans Affairs Medical Center, Durham, NC; Nathan A. Kimbrel, Duke University School of Medicine, Duke University, Durham, NC; Durham Veterans Affairs Medical Center, Durham, NC; Daniel Jacobson, Oak Ridge National Laboratory, Computational and Predictive Biology, Oak Ridge, TN //

ABSTRACT: Suicidal behavior comprises a complex psychiatric disorder and little evidence to date has identified significant genetic risk factors for suicidality. We therefore used a cross-ancestry, genome-wide association study (GWAS) to identify the causative genetic loci underlying suicide attempts using data from the Veterans’ Affairs (VA) Million Veteran Program (MVP). We identified 14,829 cases of non-fatal and fatal suicide attempts using a combination of electronic health record data from the VA Corporate Data Warehouse (CDW), Suicide Prevention Application Network (SPAN) database, and CDW Mental Health Domain; 410,464 subjects with no history of suicidal behavior were used as controls. Using genetic variants identified by GWAS, we assigned these variants to genes using H-MAGMA, a cutting-edge genomics approach used for assigning variants to genes based on Hi-C chromatin conformation data. We then used a novel systems biology, graph-driven approach (Random Walk with Restart Filter; RWR-Filter) to determine which variants, genes and associated biological pathways were dysregulated in veterans with a history of suicide attempt. RWR-Filter accurately distinguished genes contributing to suicide attempt pathophysiology from GWAS, including causative genetic variants that did not reach traditional thresholds (e.g., p-values < 1e-5) contributing to this psychiatric disorder. Furthermore, our approach provided mechanistic insight into multiple biological pathways underlying suicidal behavior that have been implicated in other psychiatric and neurological disorders, including dopaminergic signaling, microtubule dynamics, and the SWI/SNF chromatin remodeling complex. Our RWR-Filter method combined with our GWAS results may thus provide important clinical insights into the neurobiological pathways underlying suicidal behavior and putative drug targets for combating this neuropsychiatric disorder.

Source of Funding: NIH Grants DA051913 and DA051908; DOE/VA MVP CHAMPION Grant