Abstract:Background Subtyping schizophrenia can disentangle heterogeneity and help with treatment decision- making. However, current schizophrenia subtypes have not demonstrated adequate clinical utility, limited by sample size, suboptimal clustering methods, and choice of clustering input. Polygenic scores (PGS) reflect the genetic risk of phenotypes including comorbidities and are available before treatment, making them candidate clustering input. Methods We derived PGS for schizophrenia, autism spectrum disorder, bipolar disorder type-1, depression, and intelligence in 4,915 schizophrenia cases with register linkage. We randomly divided the sample into discovery and replication partitions and applied a novel clustering workflow on both: preprocessing PGS, feature extraction with uniform manifold approximation and projection (UMAP), and clustering with density-based spatial clustering of applications with noise (DBSCAN). After replication, we re-performed clustering on the entire sample and evaluated treatment-relevant variables of medication and hospitalization (extracted from registers) across clusters. Outcomes We identified five well-replicated PGS clusters. Cluster 1 (26% of entire sample) with generally lower PGS, had the least use of antipsychotics (including clozapine), and fewer outpatient visits. Cluster 2 (48%) with generally higher PGS, especially schizophrenia PGS, had more prescriptions of antipsychotics including clozapine and longer treatment with clozapine. Each featured by specific PGS, clusters 3 (high IQ-PGS, 11%), 4 (high ASD-PGS, 8%), 5 (high BIP-PGS, 7%) showed sub-threshold level significance in the corresponding phenotypic measures but did not differ significantly in the treatment-relevant variables. Solely categorizing the patients with SCZ-PGS did not generate any significant patterns in the phenotypic and treatment-relevant variables. Interpretation The results suggest that combinations of PGS of brain disorders and traits can provide clinically relevant clusters, offering a direction for future research on schizophrenia subtyping. Future replications in independent samples are required. The workflow can be generalized to other disorders and with mechanism-informed PGS. ### Competing Interest Statement The authors have declared no competing interest. ### Funding Statement PFS was supported by the Swedish Research Council (Vetenskapsradet, award D0886501), the Horizon 2020 Program of the European Union (COSYN, RIA Grant Agreement No. 610307), and US NIMH (U01 MH109528 and R01 MH077139). LY was supported by the US NIMH R01 MH123724 and MH124873, and the European Research Council (grant agreement ID: 101042183). KK was supported by the NIMH (MH123724) and the University of Manitoba. SY was supported by Karolinska Institutet. ### Author Declarations I confirm all relevant ethical guidelines have been followed, and any necessary IRB and/or ethics committee approvals have been obtained. Yes The details of the IRB/oversight body that provided approval or exemption for the research described are given below: The research procedures were approved by ethical committees at the Karolinska Institutet with written informed consent provided by the subjects. I confirm that all necessary patient/participant consent has been obtained and the appropriate institutional forms have been archived, and that any patient/participant/sample identifiers included were not known to anyone (e.g., hospital staff, patients or participants themselves) outside the research group so cannot be used to identify individuals. Yes I understand that all clinical trials and any other prospective interventional studies must be registered with an ICMJE-approved registry, such as ClinicalTrials.gov. I confirm that any such study reported in the manuscript has been registered and the trial registration ID is provided (note: if posting a prospective study registered retrospectively, please provide a statement in the trial ID field explaining why the study was not registered in advance). Yes I have followed all appropriate research reporting guidelines, such as any relevant EQUATOR Network research reporting checklist(s) and other pertinent material, if applicable. Yes All data produced in the present work are contained in the manuscript and supplementary material

Identification of transdiagnostic psychiatric disorder subtypes using unsupervised learning

Polygenic Burden Analysis of Longitudinal Clusters of Psychopathological Features in A Cross-Diagnostic Group of Individuals with Severe Mental Illness

Neurobiologically Based Stratification of Recent- Onset Depression and Psychosis: Identification of Two Distinct Transdiagnostic Phenotypes

Integrating Polygenic Allele Burden Information and Phenomic Data to Characterize Complex Disease Trajectories in Severe Mental Illness

Subtyping Schizophrenia Using Psychiatric Polygenic Scores

Identification of psychiatric disorder subtypes from functional connectivity patterns in resting-state electroencephalography

Biological Subtyping of Psychiatric Syndromes As a Pathway for Advances in Drug Discovery and Personalized Medicine

A framework for a brain-derived nosology of psychiatric disorders

Neuropsychiatric Disorder Subtyping Via Clustered Deep Learning Classifier Explanations

Two neurostructural subtypes: results of machine learning on brain images from 4,291 individuals with schizophrenia

Clinical and Biological Stratification in 121,560 Antidepressant Prescription Trajectories using Unsupervised Modelling and Clustering

Individualized gray matter morphological abnormalities uncover two robust transdiagnostic biotypes

Unique genetic and risk-factor profiles in clusters of major depressive disorder-related multimorbidity trajectories

Heterogeneity and Classification of Recent Onset Psychosis and Depression: A Multimodal Machine Learning Approach

Identifying Subgroups of Major Depressive Disorder Using Brain Structural Covariance Networks and Mapping of Associated Clinical and Cognitive Variables

Psychosis Brain Subtypes Validated in First-Episode Cohorts and Related to Illness Remission: Results from the PHENOM Consortium.

Neurostructural subgroup in 4291 individuals with schizophrenia identified using the subtype and stage inference algorithm

Two Distinct Neuroanatomical Subtypes of Schizophrenia Revealed Using Machine Learning.

Functional MRI applications for psychiatric disease subtyping: a review

Identification of neurobehavioural symptom groups based on shared brain mechanisms