No packages match

iPEB - Improved Parametric Empirical Bayes for Longitudinal Biomarker Analysis

Extends parametric empirical Bayes (PEB) for longitudinal biomarker screening with a time-gap-aware standardization layer, covariate adjustment, and objective-driven multi-marker weighting. The layer models each subject's biomarker history with a random intercept (and an optional random slope) and autocorrelated, gap-scaled residuals, so that prediction uncertainty grows with the time between visits and per-visit specificity is preserved under irregular sampling. Marker weights are learned to optimize a user-selected clinical objective -- maximizing sensitivity at a fixed specificity, extending detection lead time, or a combined objective -- with optional feature selection and a choice of scalar or multivariate combiner. Functions for fitting, prediction, and evaluation (sensitivity, lead time, and specificity at chosen operating points) are provided. A manuscript describing the method is in preparation.

Last updated

4.48 score 108 downloads

MR.RGM - Fitting Multivariate Bidirectional Mendelian Randomization Networks Using Bayesian Directed Cyclic Graphical Models

Addressing a central challenge encountered in Mendelian randomization (MR) studies, where MR primarily focuses on discerning the effects of individual exposures on specific outcomes and establishes causal links between them. Using a network-based methodology, the intricacy involving interdependent outcomes due to numerous factors has been tackled through this routine. Based on Ni et al. (2018) <doi:10.1214/17-BA1087>, 'MR.RGM' extends to a broader exploration of the causal landscape by leveraging on network structures and involves the construction of causal graphs that capture interactions between response variables and consequently between responses and instrument variables. The resulting Graph visually represents these causal connections, showing directed edges with effect sizes labeled. 'MR.RGM' facilitates the navigation of various data availability scenarios effectively by accommodating three input formats, i.e., individual-level data and two types of summary-level data. The method also optionally incorporates measured covariates (when available) and allows flexible modeling of the error variance structure, including correlated errors that may reflect unmeasured confounding among responses. In the process, causal effects, adjacency matrices, and other essential parameters of the complex biological networks, are estimated. Besides, 'MR.RGM' provides uncertainty quantification for specific network structures among response variables. Parts of the Inverse Wishart sampler are adapted from the econ722 repository by DiTraglia (GPL-2.0).

Last updated

openblascppopenmp

3.56 score 1 stars 12 scripts 298 downloads

MRCCC - Bayesian Mendelian Randomization for Causal Cell-Cell Communication

Implements MR-CCC, a Bayesian Mendelian randomization framework for inferring causal ligand-receptor signalling between cell types from population-scale single-cell expression and genotype data. For each ligand-receptor-pathway triplet, cis-eQTL genotypes instrument ligand expression in the sender cell type and receptor expression in the receiver cell type, and a working model with a ligand-by-receptor interaction is fitted to downstream pathway activity by a blocked Gibbs sampler. A spike-and-slab prior on the causal block yields a posterior inclusion probability that serves as the communication score, and the retained draws support credible intervals, sign probabilities, the receptor level at which the ligand effect reverses sign, and Bayesian false discovery rate control across many triplets.

Last updated

softwarestatisticalmethodbayesiangeneexpressionsinglecellgeneticssystemsbiologygeneregulationopenblascpp

3.00 score