Dimensional reconstruction of psychotic disorders through multimodal genetic-neural profiles
Epidemiology and Medical Biometry/Statistics
Final Report Abstract
In this reporting period, we continued with our extensive development of computational tools for the machine learning analysis of high-dimensional data, and deployed these methods to gain insight into the biology of schizophrenia. Using biologically-informed machine learning, we identified a reproducible DNA methylation signature in the blood and brain of patients with schizophrenia. We demonstrated that this signature predicted inter-individual differences in the brain-functional connectivity between the dorsolateral-prefrontal cortex (DLPFC) and the hippocampus, a robust neural intermediate phenotype of schizophrenia. Furthermore, we demonstrated that predictions from this signature were not changed in unaffected, first degree relatives, not associated with genetic risk for schizophrenia, and not altered in relevant differential diagnoses. This indicates that the identified DNA methylation signature likely reflected a schizophrenia-specific exposure to environmental risk, and was not a secondary consequence of genetic predisposition. In parallel, we implemented a computational pipeline to characterize the coordination of DNA methylation between gene-pairs at the genome wide level. Such coordination may be disturbed by environmental risk exposure, with downstream effects on gene expression and, potentially, illness susceptibility. We were able to identify a large-scale, reproducible DNA-methylation coordination network, which was particularly dense (i.e. strongly connected) in biological processes that included synaptic function. We observed that DNA methylation in these processes along the coordinated dimension was significantly altered in the DLPFC of patients with schizophrenia. Furthermore, the variance of this coordinated, synaptic methylation showed a reproducible, age-depended increase in variance. This may suggest that the underlying regulatory system is more tightly controlled in younger individuals, and it is interesting to hypothesize that environmental risk exposure may have a more pronounced functional consequence during such age periods. In line with this hypothesis, we observed that coordinated synaptic methylation showed an age dependent association with schizophrenia-relevant, brain-functional connectivity. This provides the groundwork for interesting follow-up analyses of how risk exposure mediates susceptibility via changes in DNA methylation. Finally, we expanded the development of multitask learning approaches to allow their application on geographically distributed databases, and made the developed framework publicly available.
Publications
- Protein Interaction Networks Link Schizophrenia Risk Loci to Synaptic Function. Schizophr Bull. 2016 Nov;42(6):1334-1342
Schwarz E, Izmailov R, Liò P, Meyer-Lindenberg A
(See online at https://doi.org/10.1093/schbul/sbw035) - A polygenic score for schizophrenia predicts glycemic control. Transl Psychiatry 2017; 7(12):1295
Cao H, Chen J, Meyer-Lindenberg A, Schwarz E
(See online at https://doi.org/10.1038/s41398-017-0044-z) - Gimpute: An efficient genetic data imputation pipeline. Bioinformatics, 2018;35(8):1433-1435
Chen J, Lippold D, Frank J, Rayner W, Meyer-Lindenberg A, Schwarz E
(See online at https://doi.org/10.1093/bioinformatics/bty814) - Male increase in brain gene expression variability is linked to genetic risk for schizophrenia. Transl. Psych. 2018, 8: article 140
Chen J, Cao H, Meyer-Lindenberg A, Schwarz E
(See online at https://doi.org/10.1038/s41398-018-0200-0) - Reproducible grey matter patterns index a multivariate, global alteration of brain structure in schizophrenia and bipolar disorder. Transl. Psych. 2019, 9(1):12
Schwarz E, Doan NT, Pergola G, et al.
(See online at https://doi.org/10.1038/s41398-018-0225-4) - Association of a Reproducible Epigenetic Risk Profile for Schizophrenia With Brain Methylation and Function. JAMA Psychiatry. 2020; 77(6):628-636
Chen J, Zang Z, Braun U, Schwarz K, Harneit A, Kremer T, Ma R, Schweiger J, Moessnang C, Geiger L, Cao H, Degenhardt F, Nöthen MM, Tost H, Meyer-Lindenberg A, Schwarz E
(See online at https://doi.org/10.1001/jamapsychiatry.2019.4792) - Identification of Reproducible BCL11A Alterations in Schizophrenia Through Individual-Level Prediction of Coexpression, Schizophrenia Bulletin, 2020; 46(5):1165-1171
Chen J, Cao H, Kaufmann T, T Westlye LT, Tost H, Meyer-Lindenberg A, Schwarz E
(See online at https://doi.org/10.1093/schbul/sbaa047) - Hyper-Coordinated DNA Methylation is Altered in Schizophrenia and Associated with Brain Function, Schizophrenia Bulletin Open, 2021; 2(1), sgab036
Chen J, Schwarz K, Zang Z, Braun U, Harneit A, Kremer T, Ma R, Schweiger J, Moessnang C, Geiger L, Cao H, Degenhardt F, M Nöthen MM, Tost H, Meyer-Lindenberg A, Schwarz E
(See online at https://doi.org/10.1093/schizbullopen/sgab036) - dsMTL: a computational framework for privacy-preserving, distributed multi-task machine learning. Bioinformatics. 2022; 38(21):4919-4926
Cao H, Zhang Y, Baumbach J, Burton PR, Dwyer D, Koutsouleris N, Matschinske J, Marcon Y, Rajan S, Rieg T, Ryser-Welch P, Späth J; COMMITMENT Consortium, Herrmann C, Schwarz E
(See online at https://doi.org/10.1093/bioinformatics/btac616) - From mechanistic insight towards clinical implementation using normative modelling. 2022, Nature Computational Science, 2:278–280
Rieg T, Schwarz E
(See online at https://doi.org/10.1038/s43588-022-00248-7)
