048_ Multi-level integrative omics to identify biomarkers in schizophrenia and other major psychoses (MulioBio): Classification based on psychopathology and cognitive profile
Research Question and Aims
Despite many years of research and many promising candidates there are no validated and reliable biomarkers (BM) in clinical use for schizophrenia or other major psychoses (MP). In the MulioBio project (see e.g. proposal 024_ Preparatory work for the MulioBio project and 037_ Multi-level integrative omics to identify biomarkers in a Schizophrenia and other major psychoses (MulioBio): SmallRNAome comparison of broad diagnostic groups), we postulate that phenotype definition is the reason BM could not be discovered. To resolve this, we intended to classify patients of the longitudinal PsyCourse cohort, that show the same behavioral profiles over time, into transdiagnostic groups. In the MulioBio study, former PsyCourse participants will be re-contacted, PBMCs will be obtained from them under the auspices of the MulioBio study, and a multi-level BM screening of molecular phenotypes will be performed (RNAome, smallRNAome, epigenome, proteome), eventually leading to potential biomarkers able to differentiate between the transdiagnostic groups.
The idea of the MulioBio project is that while the overlap of symptoms makes it difficult to identify biomarkers of specific disorders, it should be possible to identify biomarkers of psychosis subgroups with a similar course over time. To achieve this we will use the longmixr package (https://github.com/cellmapslab/longmixr, previously known as PhenEndo) to classify PsyCourse participants according to both the course of their psychopathology and their cognitive profile. This package realizes a clustering using longitudinal data.
We intent to use data from the 4 visits of the PsyCourse study to create transdiagnostic groups using the longmixr R package adjusting for age at baseline.
We will further characterize each group based on both phenotypic cross-sectional data (see phenotypic variables below) and PRS for schizophrenia1, bipolar disorder2, depression3 as well as a transdiagnostic PRS (calculated using PRS-CS with the summary statistics of the largest GWAS to date). To do so, we compare them using ANOVAs, t-tests, ChiČ tests, or model comparisons depending on the number of clusters and the respective variables.
The groups and the data used to generate them will eventually also be used for the analysis of newly collected data from the MulioBio project.
To characterize the cluster:
PRS-CS for schizophrenia, bipolar disorder, depression
The transdiagnostic PRS will be obtained by subjecting the disorder specific PRS to a principal component analysis, and extracting the first principal component.