Simultaneous parameter estimation and variable selection via the logit-normal continuous analogue of the spike-and-slab prior

Development and application of an optimised Bayesian shrinkage prior for spectroscopic biomedical diagnostics

BACKGROUND AND OBJECTIVE

Classification of vibrational spectra is often challenging for biological substances containing similar molecular bonds, interfering with spectral outputs. To address this, various approaches are widely studied. However, whilst providing powerful estimations, these techniques are computationally extensive and frequently overfit the data. Shrinkage priors, which favour models with relatively few predictor variables, are often applied in Bayesian penalisation techniques to avoid overfitting.

METHODS

Using the logit-normal continuous analogue of the spike-and-slab (LN-CASS) as the shrinkage prior and modelling, we have established classification for accurate analysis, with the established system found to be faster than conventional least absolute shrinkage and selection operator, horseshoe or spike-and-slab. These were examined versus coefficient data based on a linear regression model and vibrational spectra produced via density functional theory calculations. Then applied to Raman spectra from saliva to classify the sample sex.

RESULTS

Subsequently applied to the acquired spectra from saliva, the evaluated models exhibited high accuracy (AUC>90 %) even when number of parameters was higher than the number of observations. Analyses of spectra for all Bayesian models yielded high-classification accuracy upon cross-validation. Further, for saliva sensing, LN-CASS was found to be the only classifier with 100 %-accuracy in predicting the output based on a leave-one-out cross validation.

CONCLUSIONS

With potential applications in aiding diagnosis from small spectroscopic datasets and are compatible with a range of spectroscopic data formats. As seen with the classification of IR and Raman spectra. These results are highly promising for emerging developments of spectroscopic platforms for biomedical diagnostic sensing systems.

Decreased left heart flow in fetal lambs causes left heart hypoplasia and pro-fibrotic tissue remodeling

Low blood flow through the fetal left heart is often conjectured as an etiology for hypoplastic left heart syndrome (HLHS). To investigate if a decrease in left heart flow results in growth failure, we generate left ventricular inflow obstruction (LVIO) in mid-gestation fetal lambs by implanting coils in their left atrium using an ultrasound-guided percutaneous technique. Significant LVIO recapitulates important clinical features of HLHS: decreased antegrade aortic valve flow, compensatory retrograde perfusion of the brain and ascending aorta (AAo) from the arterial duct, severe left heart hypoplasia, a non-apex forming LV, and a thickened endocardial layer. The hypoplastic AAo have miRNA-gene pairs annotating to cell proliferation that are inversely differentially expressed by bulk RNA-seq. Single-nucleus RNA-seq of the hypoplastic LV myocardium shows an increase in fibroblasts with a reciprocal decrease in cardiomyocyte nuclei proportions. Fibroblasts, cardiomyocytes and endothelial cells from hypoplastic myocardium have increased expression of extracellular matrix component or fibrosis genes with dysregulated fibroblast growth factor signaling. Hence, a severe sustained ( ~ 1/3 gestation) reduction in fetal left heart flow is sufficient to cause left heart hypoplasia. This is accompanied by changes in cellular composition and gene expression consistent with a pro-fibrotic environment and aberrant induction of mesenchymal programs.

Integrating single-cell sequencing data with GWAS summary statistics reveals CD16+monocytes and memory CD8+T cells involved in severe COVID-19

Background

Understanding the host genetic architecture and viral immunity contributes to the development of effective vaccines and therapeutics for controlling the COVID-19 pandemic. Alterations of immune responses in peripheral blood mononuclear cells play a crucial role in the detrimental progression of COVID-19. However, the effects of host genetic factors on immune responses for severe COVID-19 remain largely unknown.

Methods

We constructed a computational framework to characterize the host genetics that influence immune cell subpopulations for severe COVID-19 by integrating GWAS summary statistics (N = 969,689 samples) with four independent scRNA-seq datasets containing healthy controls and patients with mild, moderate, and severe symptom (N = 606,534 cells). We collected 10 predefined gene sets including inflammatory and cytokine genes to calculate cell state score for evaluating the immunological features of individual immune cells.

Results

We found that 34 risk genes were significantly associated with severe COVID-19, and the number of highly expressed genes increased with the severity of COVID-19. Three cell subtypes that are CD16+monocytes, megakaryocytes, and memory CD8+T cells were significantly enriched by COVID-19-related genetic association signals. Notably, three causal risk genes of CCR1, CXCR6, and ABO were highly expressed in these three cell types, respectively. CCR1⁺CD16+monocytes and ABO⁺ megakaryocytes with significantly up-regulated genes, including S100A12, S100A8, S100A9, and IFITM1, confer higher risk to the dysregulated immune response among severe patients. CXCR6⁺ memory CD8+ T cells exhibit a notable polyfunctionality including elevation of proliferation, migration, and chemotaxis. Moreover, we observed an increase in cell-cell interactions of both CCR1⁺ CD16+monocytes and CXCR6⁺ memory CD8+T cells in severe patients compared to normal controls among both PBMCs and lung tissues. The enhanced interactions of CXCR6⁺ memory CD8+T cells with epithelial cells facilitate the recruitment of this specific population of T cells to airways, promoting CD8+T cell-mediated immunity against COVID-19 infection.

Conclusions

We uncover a major genetics-modulated immunological shift between mild and severe infection, including an elevated expression of genetics-risk genes, increase in inflammatory cytokines, and of functional immune cell subsets aggravating disease severity, which provides novel insights into parsing the host genetic determinants that influence peripheral immune cells in severe COVID-19.

Supplementary Information

The online version contains supplementary material available at 10.1186/s13073-022-01021-1.

scCODA is a Bayesian model for compositional single-cell data analysis

Compositional changes of cell types are main drivers of biological processes. Their detection through single-cell experiments is difficult due to the compositionality of the data and low sample sizes. We introduce scCODA ( https://github.com/theislab/scCODA ), a Bayesian model addressing these issues enabling the study of complex cell type effects in disease, and other stimuli. scCODA demonstrated excellent detection performance, while reliably controlling for false discoveries, and identified experimentally verified cell type changes that were missed in original analyses.

Steroid Metabolome Analysis in Disorders of Adrenal Steroid Biosynthesis and Metabolism

Steroid biosynthesis and metabolism are reflected by the serum steroid metabolome and, in even more detail, by the 24-hour urine steroid metabolome, which can provide unique insights into alterations of steroid flow and output indicative of underlying conditions. Mass spectrometry-based steroid metabolome profiling has allowed for the identification of unique multisteroid signatures associated with disorders of steroid biosynthesis and metabolism that can be used for personalized approaches to diagnosis, differential diagnosis, and prognostic prediction. Additionally, steroid metabolome analysis has been used successfully as a discovery tool, for the identification of novel steroidogenic disorders and pathways as well as revealing insights into the pathophysiology of adrenal disease. Increased availability and technological advances in mass spectrometry-based methodologies have refocused attention on steroid metabolome profiling and facilitated the development of high-throughput steroid profiling methods soon to reach clinical practice. Furthermore, steroid metabolomics, the combination of mass spectrometry-based steroid analysis with machine learning-based approaches, has facilitated the development of powerful customized diagnostic approaches. In this review, we provide a comprehensive up-to-date overview of the utility of steroid metabolome analysis for the diagnosis and management of inborn disorders of steroidogenesis and autonomous adrenal steroid excess in the context of adrenal tumors.