Jointly achieving parsimony and good predictive power in high dimensions is a main challenge in statistics. Non-local priors (NLPs) possess appealing properties for model choice, but their use for estimation has not been studied in detail. We show that for regular models NLP-based Bayesian model averaging (BMA) shrink spurious parameters either at fast polynomial or quasi-exponential rates as the sample size n increases, while non-spurious parameter estimates are not shrunk. We extend some results to linear models with dimension p growing with n. Coupled with our theoretical investigations, we outline the constructive representation of NLPs as mixtures of truncated distributions that enables simple posterior sampling and extending NLPs beyond previous proposals. Our results show notable high-dimensional estimation for linear models with p ≫ n at low computational cost. NLPs provided lower estimation error than benchmark and hyper-g priors, SCAD and LASSO in simulations, and in gene expression data achieved higher cross-validated R2 with less predictors. Remarkably, these results were obtained without pre-screening variables. Our findings contribute to the debate of whether different priors should be used for estimation and model selection, showing that selection priors may actually be desirable for high-dimensional estimation.

MHC-I molecules expose the intracellular protein content on the cell surface, allowing T cells to detect foreign or mutated peptides. The combination of six MHC-I alleles each individual carries defines the sub-peptidome that can be effectively presented. We applied this concept to human cancer, hypothesizing that oncogenic mutations could arise in gaps in personal MHC-I presentation. To validate this hypothesis, we developed and applied a residue-centric patient presentation score to 9,176 cancer patients across 1,018 recurrent oncogenic mutations. We found that patient MHC-I genotype-based scores could predict which mutations were more likely to emerge in their tumor. Accordingly, poor presentation of a mutation across patients was correlated with higher frequency among tumors. These results support that MHC-I genotype-restricted immunoediting during tumor formation shapes the landscape of oncogenic mutations observed in clinically diagnosed tumors and paves the way for predicting personal cancer susceptibilities from knowledge of MHC-I genotype.

Compensatory proliferation triggered by hepatocyte loss is required for liver regeneration and maintenance but also promotes development of hepatocellular carcinoma (HCC). Despite extensive investigation, the cells responsible for hepatocyte restoration or HCC development remain poorly characterized. We used genetic lineage tracing to identify cells responsible for hepatocyte replenishment following chronic liver injury and queried their roles in three distinct HCC models. We found that a pre-existing population of periportal hepatocytes, located in the portal triads of healthy livers and expressing low amounts of Sox9 and other bile-duct-enriched genes, undergo extensive proliferation and replenish liver mass after chronic hepatocyte-depleting injuries. Despite their high regenerative potential, these so-called hybrid hepatocytes do not give rise to HCC in chronically injured livers and thus represent a unique way to restore tissue function and avoid tumorigenesis. This specialized set of pre-existing differentiated cells may be highly suitable for cell-based therapy of chronic hepatocyte-depleting disorders.

MHC-I molecules expose the intracellular protein content on the cell surface, allowing T cells to detect foreign or mutated peptides. The combination of six MHC-I alleles each individual carries defines the sub-peptidome that can be effectively presented. We applied this concept to human cancer, hypothesizing that oncogenic mutations could arise in gaps in personal MHC-I presentation. To validate this hypothesis, we developed and applied a residue-centric patient presentation score to 9,176 cancer patients across 1,018 recurrent oncogenic mutations. We found that patient MHC-I genotype-based scores could predict which mutations were more likely to emerge in their tumor. Accordingly, poor presentation of a mutation across patients was correlated with higher frequency among tumors. These results support that MHC-I genotype-restricted immunoediting during tumor formation shapes the landscape of oncogenic mutations observed in clinically diagnosed tumors and paves the way for predicting personal cancer susceptibilities from knowledge of MHC-I genotype.