A primer on the use of machine learning to distil knowledge from data in biological psychiatry.

Applications of machine learning in the biomedical sciences are growing rapidly. This growth has been spurred by diverse cross-institutional and interdisciplinary collaborations, public availability of large datasets, an increase in the accessibility of analytic routines, and the availability of powerful computing resources. With this increased access and exposure to machine learning comes a responsibility for education and a deeper understanding of its bases and bounds, borne equally by data scientists seeking to ply their analytic wares in medical research and by biomedical scientists seeking to harness such methods to glean knowledge from data. This article provides an accessible and critical review of machine learning for a biomedically informed audience, as well as its applications in psychiatry. The review covers definitions and expositions of commonly used machine learning methods, and historical trends of their use in psychiatry. We also provide a set of standards, namely Guidelines for REporting Machine Learning Investigations in Neuropsychiatry (GREMLIN), for designing and reporting studies that use machine learning as a primary data-analysis approach. Lastly, we propose the establishment of the Machine Learning in Psychiatry (MLPsych) Consortium, enumerate its objectives, and identify areas of opportunity for future applications of machine learning in biological psychiatry. This review serves as a cautiously optimistic primer on machine learning for those on the precipice as they prepare to dive into the field, either as methodological practitioners or well-informed consumers.

microRNA and the Post-Transcriptional Response to Oxidative Stress during Neuronal Differentiation: Implications for Neurodevelopmental and Psychiatric Disorders.

Oxidative stress is one of the most important environmental exposures associated with psychiatric disorders, but the underlying molecular mechanisms remain to be elucidated. In a previous study, we observed a substantial alteration of the gene expression landscape in neuron-like cells that were differentiated from SH-SY5Y cells after or during exposure to oxidative stress, with a subset of dysregulated genes being enriched for neurodevelopmental processes. To further explore the regulatory mechanisms that might account for such profound perturbations, we have now applied small RNA-sequencing to investigate changes in the expression of miRNAs. These molecules are known to play crucial roles in brain development and response to stress through their capacity to suppress gene expression and influence complex biological networks. Through these analyses, we observed more than a hundred differentially expressed miRNAs, including 80 previously reported to be dysregulated in psychiatric disorders. The seven most influential miRNAs associated with pre-treatment exposure, including miR-138-5p, miR-96-5p, miR-34c-5p, miR-1287-5p, miR-497-5p, miR-195-5p, and miR-16-5p, supported by at least 10 negatively correlated mRNA connections, formed hubs in the interaction network with 134 genes enriched with neurobiological function, whereas in the co-treatment condition, miRNA-mRNA interaction pairs were enriched in cardiovascular and immunity-related disease ontologies. Interestingly, 12 differentially expressed miRNAs originated from the DLK1-DIO3 location, which encodes a schizophrenia-associated miRNA signature. Collectively, our findings suggest that early exposure to oxidative stress, before and during prenatal neuronal differentiation, might increase the risk of mental illnesses in adulthood by disturbing the expression of miRNAs that regulate neurodevelopmentally significant genes and networks.

miRNA cargo in circulating vesicles from neurons is altered in individuals with schizophrenia and associated with severe disease.

While RNA expression appears to be altered in several brain disorders, the constraints of postmortem analysis make it impractical for well-powered population studies and biomarker development. Given that the unique molecular composition of neurons are reflected in their extracellular vesicles (EVs), we hypothesized that the fractionation of neuron derived EVs provides an opportunity to specifically profile their encapsulated contents noninvasively from blood. To investigate this hypothesis, we determined miRNA expression in microtubule associated protein 1B (MAP1B)-enriched serum EVs derived from neurons from a large cohort of individuals with schizophrenia and nonpsychiatric comparison participants. We observed dysregulation of miRNA in schizophrenia subjects, in particular those with treatment-resistance and severe cognitive deficits. These data support the hypothesis that schizophrenia is associated with alterations in posttranscriptional regulation of synaptic gene expression and provides an example of the potential utility of tissue-specific EV analysis in brain disorders.

BDNF and the maturation of posttranscriptional regulatory networks in human SH-SY5Y neuroblast differentiation.

The SH-SY5Y culture system is a convenient neuronal model with the potential to elaborate human/primate-specific transcription networks and pathways related to human cognitive disorders. While this system allows for the exploration of specialized features in the human genome, there is still significant debate about how this model should be implemented, and its appropriateness for answering complex functional questions related to human neural architecture. In view of these questions we sought to characterize the posttranscriptional regulatory structure of the two-stage ATRA differentiation, BDNF maturation protocol proposed by Encinas et al. (2000) using integrative whole-genome gene and microRNA (miRNA) expression analysis. We report that ATRA-BDNF induced significant increases in expression of key synaptic genes, brain-specific miRNA and miRNA biogenesis machinery, and in AChE activity, compared with ATRA alone. Functional annotation clustering associated BDNF more significantly with neuronal terms, and with synaptic terms not found in ATRA-only clusters. While our results support use of SH-SY5Y as a neuronal model, we advocate considered selection of the differentiation agent/s relative to the system being modeled.