031_ Assessment of LipidQTLs in the PsyCourse Study
Research Question and Aims
As the most abundant compounds in the central nervous system, lipids are known to play an important yet often times disregarded role in neuronal structure and function. This role includes tasks in the regulation of membrane fluidity and permeability, vesicle formation and transport, neurotransmitter release, cell integrity and plasticity. Dysfunction of these processes has been implicated in the pathogenesis of numerous psychiatric disorders (reviewed in PMID 27054615).
There is mounting evidence that lipid species are altered both in the central nervous system and the periphery of individuals with severe psychiatric disorders such as bipolar disorder (BPD) or schizophrenia (SCZ) compared to controls (PMID 30089790, own unpublished observation). As comprehensive lipidomics are a still-emerging field, few studies have assessed the genetic underpinnings of lipid profiles in a comprehensive fashion. A recent study performed a plasma lipid species-based quantitative trait locus (lipidQTL) analysis and identified 42 genome-wide significant associations between single nucleotide polymorphisms (SNPs) and molecular lipid species (PMID 31551469). Interestingly, 19 of the proposed gene candidates at these lipid-species-associated loci also harbor common genetic variants associated with psychiatric phenotypes ranging from cognitive phenotypes and major psychiatric disorders such as BPD or SCZ to transdiagnostic phenotypes (cf. GWAS Catalog, www.ebi.ac.uk/gwas/).
Accordingly, herein, we propose to perform a lipidQTL analysis, individually or as a low-dimensional representation of the lipidomic and proteomic datasets, in plasma samples of individuals belonging to the PsyCourse Study for whom plasma lipidomic profiles (n=310 BPD, n=234 SCZ, n=266 controls) and genotype data are available. Aim one would be to replicate the findings of the performed lipidQTL study above in an independent data set. The second aim would be to use lipidQTLs to identify novel pathways or pathomechanisms involved in BPD and SCZ.
We postulate that plasma lipidQTLs may represent a tool to uncover pathomechanisms underlying BPD and SCZ.
We will use SCID-assessed DSM-IV diagnoses for phenotype definition.
Available lipidomics data have been quality controlled by the collaborating lab, quality control of genotype data will performed as described previously in studies in which the applicants were involved (e.g. PMID 26806518, PMID 31712617).
LipidQTL analysis will be performed using Matrix eQTL for each lipid species available in our sample (http://www.bios.unc.edu/research/ genomic_software/Matrix_eQTL/runit.html).
In a complementary approach, we will use Multi-Omics Factor Analysis v2 (MOFA+) to generate an interpretable low-dimensional representation in terms of a few latent factors regarding lipidomics or (lipidomics + proteomics). Subsequently these dimensions will be used as target variables in QTL analyses using PLINK1.90.
PCA will be performed to address potential confounders.
The identified lipidQTLs will be blasted against large databases of know genetic risk loci not only for mental disorders but many other human traits with the aim to identify genetic overlaps. Moreover, depending on the outcomes, a phenome-wide analysis of the top variants on selected phenotypes will be carried out using resources like the UK Biobank.
raw medication data sets (v1_med_clin_orig, v1_med_con_orig)
raw medication data sets (v2_med_clin_orig, v2_med_con_orig)
raw medication data sets (v3_med_clin_orig, v3_med_con_orig)
raw medication data sets (v4_med_clin_orig, v4_med_con_orig)
Raw genotypes to calculate PCAs
Imputed genotypes for the calculation of PRS