The hippocampus is a brain structure known for its role in declarative memory- our ability to consciously recall facts and events. The hippocampus is a highly heterogeneous brain structure, and the small subregion CA2 has been shown to be necessary for the formation of social memories, the ability of an animal to recognize previously encountered conspecifics. Changes in excitatory/inhibitory balance have been observed in CA2 in humans with schizophrenia and in mouse models of schizophrenia, suggesting that these alterations may lead to some of the social dysfunction seen in schizophrenia. Although the hippocampal CA2 region has been implicated in social memory and neuropsychiatric disorders, little is known about how CA2 neural activity may encode social interactions and how this coding may be altered in disease. To see if and how CA2 codes for social interactions, I recorded extracellularly from CA2 pyramidal neurons as mice engage in a three-chamber social interaction task where the mice interact with the following task dimensions: space, novel objects, familiar social stimuli, novel social stimuli, and the passage of time. I found that whereas CA2 activity fails to provide a stable representation of space, unlike most other dorsal hippocampal subregions, it does code for contextual changes and for novel social stimuli. In Df(16)A+/- mice, which model the 22q11.2 microdeletion, a major schizophrenia risk factor, CA2 activity fails to encode context or social novelty, consistent with the deficit in social memory seen in these mice. In contrast, CA2 activity shows a surprising increase in spatial coding in Df(16)A+/- mice. These mice were previously shown to have a loss of inhibitory neurons within CA2, and a hyperpolarization of the CA2 pyramidal neuron resting potential. This hyperpolarization is likely due to upregulation of the outward rectifying TREK-1 K+ channel. I found that administration of a TREK-1 K+ channel antagonist rescued social memory and restored the normal CA2 coding properties in the mutants. These results demonstrate a crucial role for CA2 in the encoding of social stimuli and the expression of social memory, and suggest that dysfunction in CA2 may underlie deficits in social function seen in some forms of neuropsychiatric disease.

Early life adversity (ELA) produces long lasting developmental changes to the postnatal brain, increasing predisposition to a number of physical and psychiatric disorders. The mechanisms through which ELA is able to create lasting detrimental changes to neuronal development remains unclear. This thesis tested the hypothesis that increases in fear generalization, a common symptom in psychiatric disorders, follows ELA exposure in age dependent and sexually dimorphic ways in alignment with the findings of clinical studies. The effects of ELA often impact fear circuitry and we confirmed, using electrophysiology and tissue imaging, that 5-HT circuitry from the median raphe nucleus (MRN), integral to fear response, was impaired following ELA. Using a transgenic mouse model that allows for modulation of serotonergic release, we showed that circumventing serotonergic pathways disrupted by ELA and increasing whole brain 5-HT release was enough to rescue hippocampal dependent fear responses and fear generalization. Involvement of the hippocampus in ELA effects, particularly the ventral dentate gyrus (vDG), in fear overgeneralization was confirmed as hyperactivity in thevDG following exposure to novel contexts was rescued by increased 5-HT release. In addition to ELA-induced hyperactivity of the vDG, known to potentiate stress susceptibility, I demonstrated that ELA resulted in an increase in passive coping strategies, HPA axis dysfunction and elevated stress hormone release. These effects were seen predominantly in adult females and rescued in those with increased 5-HT release. Together these data suggest that increased predisposition to psychiatric disorders following ELA exposure involves the disruption of fear circuitry regulated by 5-HT activity. Identifying the underlying circuits altered by ELA not only provides insight about disrupted postnatal brain development, but also increases our knowledge of the timeline, trajectory and factors affecting healthy postnatal brain development.

The overarching goal of this dissertation is to integrate heterogeneous data for the estimation of disease coheritability and subtyping. Chapter 2 focuses on the significance and estimation of heritability and coheritability, which quantify the proportion of phenotypic variation attributable to genetic factors and the genetic correlations between different traits, respectively. To achieve this, we develop robust statistical methods based on estimating equations that account for familial correlations and the computational challenges posed by large pedigrees and extensive datasets. Our methods are evaluated through simulations, demonstrating satisfactory consistency and robust inference properties. Compared to simpler methods performing separate trait analysis, our approaches show a greater power through joint analysis of multiple traits. An application to the analysis of heritability and coheritability in electronic health record (EHR) data reveals substantial genetic correlations between mental disorders and metabolic/endocrine measurements, suggesting shared genetic influences that warrant further investigation. These findings have implications for understanding these conditions' etiology, diagnosis, and treatment. Chapters 3 and 4 focus on the importance of patient subtyping for personalized mental health care, particularly relevant to the substantial variability observed in mental disorders. Chapter 3 develops methods for subtyping patients with mental disorders using various data modalities and variational inference. We propose latent mixture models inspired by the Item Response Theory to handle both binary and continuous data. We also introduce Black Box Variational Inference (BBVI) algorithms to overcome the challenges of numeric integration in nonlinear models. Our numerical experiments validate the proposed methods, demonstrating that variance-controlling techniques improve convergence speed and reduce iteration variance. However, the proposed algorithm encounters limitations with latent mixture models containing binary modalities due to approximations used in non-conjugate posterior distributions resulting from the non-exponential family likelihood function. Chapter 4 investigates multi-modal integration techniques for subtyping patients using data from the Adolescent Brain Cognitive Development (ABCD) study. We introduce a Bayesian hierarchical joint model with latent variables and utilize Pólya-Gamma augmentation for posterior approximation, which enables efficient Gibbs sampling and accurate estimation of model parameters. Extensive simulations confirm the consistency of estimators and the prediction accuracy of our method. Applying these methods to patient clustering in the ABCD study provides information for identifying potential clinical subtypes within mental health, which can inform the development of targeted psychological and educational interventions, ultimately improving mental health outcomes. Keywords: latent mixture model, integrative analysis, coheritability, multi-modality, disease subtyping, variational inference, Pólya-Gamma