2019-05-13

012_ An investigation of transdiagnostic psychosis subgroups with prognostic and genetic validation

Research Question and Aims

This analysis is an addition to the paper "An investigation of transdiagnostic psychosis subgroups with prognostic and genetic validation" that was recently rejected from JAMA Psychiatry. In this paper we used clinical/cognitive variables and schizophrenia PRS in order to validate subgroups detected using a matrix factorization method. Based on reviewer comments, we want to better contextualise the results with depression and bipolar PRS in addition to education polygenic scores. The additional diagnostic scores are being used because we want to be able to explain the contribution of each to the subgroups. We want to use education because the fifth subgroup we found contains individuals with particularly poor education outcomes and we think that this could be related to an endophenotype of this clinically defined subgroup.

Analytic Plan

Hypotheses:
1. the proportion of individuals with each diagnosis will determine relationships with depression, bipolar, and schizophrenia PRS;
2. the education polygenic score will be depressed in the fifth subgroup;
We intent to conduct the analyses initially on the PsyCourse phase 1 data release where we have 765 individuals. We will then attempt to replicate the results by using 485 additional individuals from the phase 3 database. We will conduct ANCOVA analyses to determine subgroup differences across all PRS thresholds using the first 4 PCA components in order to control for ancestry effects. If there is insufficient power then we will combine the discovery and replication cohorts.

Resources needed

The PRS scores for depression, bipolar, schizophrenia, and education.