044_ Blood-based transcriptome signatures of suicidality
Research Question and Aims
Very recent data indicate that networks of gene co-expression that can be measured in the postmortem brains of neuropsychiatrically normal individuals are very well preserved in the peripheral blood;1 thus, at the level of co-expressed gene modules, the biological networks of the brain can be recapitulated from measurements made in blood, increasing the feasibility of mechanistic studies of neuropsychiatric disease. Further, we recently detected reliable within-subject relationships between principal components of peripheral-blood gene-expression levels and the expression levels of individual transcripts in the brain in the same individuals. We developed a method, called BrainGENIE (the Brain Gene Expression and Network Imputation Engine), that uses those relationships as a basis for imputing gene expression in 12 distinct brain regions based on transcriptome measures from peripheral blood. In the proposed study, we would use existing peripheral-blood-based transcriptome data (LexoGene data) from all PsyCourse subjects to identify biomarkers that correlate with state measures of suicidality, and to identify corresponding changes in 12 brain regions through imputation using BrainGENIE. In the spirit of the U.S. NIMH RDoC (Research Domain Criteria) framework, we will examine suicidality as a cross-disorder construct, so will welcome all subjects regardless of diagnosis.
The analysis will use the PsyCourse samples genotyped using the Global Screening Array (GSA). Using imputed data, polygenic risk scores for SCZ, BD, MDD, Educational Attainment and probably others based on upcoming GWASes of interest will be calculated in these subjects.
The potential effect of the interaction of PRS with Environmental factors (Childhood Trauma/Life Events) on several cross-sectional outcomes: drug use (use/not use, frequency of use) and clinical severity (as measured by psychopathological scales of positive/negative/manic/depressive symptoms) in psychiatric patients will be statistically assessed using linear/logistic models. These models will include covariates as well: age, sex, duration of disease, and ancestry components, among others. The relevance of the covariates for each model will be assessed using the AIC criteria.
Raw Medication data at each visit
Raw genotypes pre-imputation to calculate PCAs
FASTQ and BAM files of Lexogen 3' RNA