We propose an adaptive nuclear norm penalization approach for low-rank matrix approximation, and use it to develop a new reduced rank estimation method for high-dimensional multivariate regression. The adaptive nuclear norm is defined as the weighted sum of the singular values of the matrix, and it is generally nonconvex under the natural restriction that the weight decreases with the singular value. However, we show that the proposed nonconvex penalized regression method has a global optimal solution obtained from an adaptively soft-thresholded singular value decomposition. The method is computationally efficient, and the resulting solution path is continuous. The rank consistency of and prediction/estimation performance bounds for the estimator are established for a high-dimensional asymptotic regime. Simulation studies and an application in genetics demonstrate its efficacy.

Accurate prediction of suicide risk among children and adolescents within an actionable time frame is an important but challenging task. Very few studies have comprehensively considered the clinical risk factors available to produce quantifiable risk scores for estimation of short- and long-term suicide risk for pediatric population. In this paper, we built machine learning models for predicting suicidal behavior among children and adolescents based on their longitudinal clinical records, and determining short- and long-term risk factors. This retrospective study used deidentified structured electronic health records (EHR) from the Connecticut Children’s Medical Center covering the period from 1 October 2011 to 30 September 2016. Clinical records of 41,721 young patients (10–18 years old) were included for analysis. Candidate predictors included demographics, diagnosis, laboratory tests, and medications. Different prediction windows ranging from 0 to 365 days were adopted. For each prediction window, candidate predictors were first screened by univariate statistical tests, and then a predictive model was built via a sequential forward feature selection procedure. We grouped the selected predictors and estimated their contributions to risk prediction at different prediction window lengths. The developed predictive models predicted suicidal behavior across all prediction windows with AUCs varying from 0.81 to 0.86. For all prediction windows, the models detected 53–62% of suicide-positive subjects with 90% specificity. The models performed better with shorter prediction windows and predictor importance varied across prediction windows, illustrating short- and long-term risks. Our findings demonstrated that routinely collected EHRs can be used to create accurate predictive models for suicide risk among children and adolescents.

Multiple sclerosis (MS) has a complex genetic, immune and metabolic pathophysiology. Recent studies implicated the gut microbiome in MS pathogenesis. However, interactions between the microbiome and host immune system, metabolism and diet have not been studied over time in this disorder. We performed a six-month longitudinal multi-omics study of 49 participants (24 untreated relapse remitting MS patients and 25 age, sex, race matched healthy control individuals. Gut microbiome composition and function were characterized using 16S and metagenomic shotgun sequencing. Flow cytometry was used to characterize blood immune cell populations and cytokine profiles. Circulating metabolites were profiled by untargeted UPLC-MS. A four-day food diary was recorded to capture the habitual dietary pattern of study participants. Together with changes in blood immune cells, metagenomic analysis identified a number of gut microbiota decreased in MS patients compared to healthy controls, and microbiota positively or negatively correlated with degree of disability in MS patients. MS patients demonstrated perturbations of their blood metabolome, such as linoleate metabolic pathway, fatty acid biosynthesis, chalcone, dihydrochalcone, 4-nitrocatechol and methionine. Global correlations between multi-omics demonstrated a disrupted immune-microbiome relationship and a positive blood metabolome-microbiome correlation in MS. Specific feature association analysis identified a potential correlation network linking meat servings with decreased gut microbe B. thetaiotaomicron, increased Th17 cell and greater abundance of meat-associated blood metabolites. The microbiome and metabolome profiles remained stable over six months in MS and control individuals. Our study identified multi-system alterations in gut microbiota, immune and blood metabolome of MS patients at global and individual feature level. Multi-OMICS data integration deciphered a potential important biological network that links meat intakes with increased meat-associated blood metabolite, decreased polysaccharides digesting bacteria, and increased circulating proinflammatory marker. This work was supported by the Washington University in St. Louis Institute of Clinical and Translational Sciences, funded, in part, by Grant Number # UL1 TR000448 from the National Institutes of Health, National Center for Advancing Translational Sciences, Clinical and Translational Sciences Award (Zhou Y, Piccio, L, Lovett-Racke A and Tarr PI); R01 NS10263304 (Zhou Y, Piccio L); the Leon and Harriet Felman Fund for Human MS Research (Piccio L and Cross AH). Cantoni C. was supported by the National MS Society Career Transition Fellowship (TA-180531003) and by donations from Whitelaw Terry, Jr. / Valerie Terry Fund. Ghezzi L. was supported by the Italian Multiple Sclerosis Society research fellowship (FISM 2018/B/1) and the National Multiple Sclerosis Society Post-Doctoral Fellowship (FG-190734474). Anne Cross was supported by The Manny & Rosalyn Rosenthal-Dr. John L. Trotter MS Center Chair in Neuroimmunology of the Barnes-Jewish Hospital Foundation. The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health.

The lateral hypothalamic area (LHA) lies at the intersection of multiple neural and humoral systems and orchestrates fundamental aspects of behavior. Two neuronal cell types found in the LHA are defined by their expression of hypocretin/orexin (Hcrt/Ox) and melanin-concentrating hormone (MCH) and are both important regulators of arousal, feeding, and metabolism. Conflicting evidence suggests that these cell populations have a more complex signaling repertoire than previously appreciated, particularly in regard to their coexpression of other neuropeptides and the machinery for the synthesis and release of GABA and glutamate. Here, we undertook a single-cell expression profiling approach to decipher the neurochemical phenotype, and heterogeneity therein, of Hcrt/Ox and MCH neurons. In transgenic mouse lines, we used single-cell quantitative polymerase chain reaction (qPCR) to quantify the expression of 48 key genes, which include neuropeptides, fast neurotransmitter components, and other key markers, which revealed unexpected neurochemical diversity. We found that single MCH and Hcrt/Ox neurons express transcripts for multiple neuropeptides and markers of both excitatory and inhibitory fast neurotransmission. Virtually all MCH and approximately half of the Hcrt/Ox neurons sampled express both the machinery for glutamate release and GABA synthesis in the absence of a vesicular GABA release pathway. Furthermore, we found that this profile is characteristic of a subpopulation of LHA glutamatergic neurons but contrasts with a broad population of LHA GABAergic neurons. Identifying the neurochemical diversity of Hcrt/Ox and MCH neurons will further our understanding of how these populations modulate postsynaptic excitability through multiple signaling mechanisms and coordinate diverse behavioral outputs.