A natural Bayesian approach for mixture models with an unknown number of components is to take the usual finite mixture model with symmetric Dirichlet weights, and put a prior on the number of components—that is, to use a mixture of finite mixtures (MFM). The most commonly-used method of inference for MFMs is reversible jump Markov chain Monte Carlo, but it can be nontrivial to design good reversible jump moves, especially in high-dimensional spaces. Meanwhile, there are samplers for Dirichlet process mixture (DPM) models that are relatively simple and are easily adapted to new applications. It turns out that, in fact, many of the essential properties of DPMs are also exhibited by MFMs—an exchangeable partition distribution, restaurant process, random measure representation, and stick-breaking representation—and crucially, the MFM analogues are simple enough that they can be used much like the corresponding DPM properties. Consequently, many of the powerful methods developed for inference in DPMs can be directly applied to MFMs as well; this simplifies the implementation of MFMs and can substantially improve mixing. We illustrate with real and simulated data, including high-dimensional gene expression data used to discriminate cancer subtypes.

Clinically relevant subtypes exist for pancreatic ductal adenocarcinoma (PDAC), but molecular characterization is not yet standard in clinical care. We implemented a biopsy protocol to perform time-sensitive whole-exome sequencing and RNA sequencing for patients with advanced PDAC. Therapeutically relevant genomic alterations were identified in 48% (34/71) and pathogenic/likely pathogenic germline alterations in 18% (13/71) of patients. Overall, 30% (21/71) of enrolled patients experienced a change in clinical management as a result of genomic data. Twenty-six patients had germline and/or somatic alterations in DNA-damage repair genes, and 5 additional patients had mutational signatures of homologous recombination deficiency but no identified causal genomic alteration. Two patients had oncogenic in-frame deletions, and we report the first clinical evidence that this alteration confers sensitivity to MAPK pathway inhibition. Moreover, we identified tumor/stroma gene expression signatures with clinical relevance. Collectively, these data demonstrate the feasibility and value of real-time genomic characterization of advanced PDAC. Molecular analyses of metastatic PDAC tumors are challenging due to the heterogeneous cellular composition of biopsy specimens and rapid progression of the disease. Using an integrated multidisciplinary biopsy program, we demonstrate that real-time genomic characterization of advanced PDAC can identify clinically relevant alterations that inform management of this difficult disease. .

The standard approach to Bayesian inference is based on the assumption that the distribution of the data belongs to the chosen model class. However, even a small violation of this assumption can have a large impact on the outcome of a Bayesian procedure. We introduce a simple, coherent approach to Bayesian inference that improves robustness to perturbations from the model: rather than condition on the data exactly, one conditions on a neighborhood of the empirical distribution. When using neighborhoods based on relative entropy estimates, the resulting "coarsened" posterior can be approximated by simply tempering the likelihood-that is, by raising it to a fractional power-thus, inference is often easily implemented with standard methods, and one can even obtain analytical solutions when using conjugate priors. Some theoretical properties are derived, and we illustrate the approach with real and simulated data, using mixture models, autoregressive models of unknown order, and variable selection in linear regression.

Novel coronavirus disease 2019 (COVID-19) severity is highly variable, with pediatric patients typically experiencing less severe infection than adults and especially the elderly. The basis for this difference is unclear. We find that mRNA and protein expression of angiotensin-converting enzyme 2 (ACE2), the cell entry receptor for the novel severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) that causes COVID-19, increases with advancing age in distal lung epithelial cells. However, in humans, ACE2 expression exhibits high levels of intra- and interindividual heterogeneity. Further, cells infected with SARS-CoV-2 experience endoplasmic reticulum stress, triggering an unfolded protein response and caspase-mediated apoptosis, a natural host defense system that halts virion production. Apoptosis of infected cells can be selectively induced by treatment with apoptosis-modulating BH3 mimetic drugs. Notably, epithelial cells within young lungs and airways are more primed to undergo apoptosis than those in adults, which may naturally hinder virion production and support milder COVID-19 severity.

The uniform distribution on matrices with specified row and column sums is often a natural choice of null model when testing for structure in two-way tables (binary or nonnegative integer). Due to the difficulty of sampling from this distribution, many approximate methods have been developed. We will show that by exploiting certain symmetries, exact sampling and counting is in fact possible in many nontrivial real-world cases. We illustrate with real datasets including ecological co-occurrence matrices and contingency tables.Comment: Published in at http://dx.doi.org/10.1214/13-AOS1131 the Annals of Statistics (http://www.imstat.org/aos/) by the Institute of Mathematical Statistics (http://www.imstat.org). arXiv admin note: text overlap with arXiv:1104.032