Immunotherapy has revolutionized cancer treatment, but predicting patient response remains challenging. Here, we presented Intelligent Predicting Response to cancer Immunotherapy through Systematic Modeling (iPRISM), a novel network-based model that integrates multiple data types to predict immunotherapy outcomes. It incorporates gene expression, biological functional network, tumor microenvironment characteristics, immune-related pathways, and clinical data to provide a comprehensive view of factors influencing immunotherapy efficacy. By identifying key genetic and immunological factors, it provides an insight for more personalized treatment strategies and combination therapies to overcome resistance mechanisms.

Fits Bayesian hierarchical models of vaccine coverage by location, birth cohort, and age. Observations of vaccination status (which may span cohorts, ages, and doses) are used to fit a latent survival-style process model that decomposes coverage into a lifetime propensity to vaccinate and a time-varying force of vaccination. Hierarchical spatial structure (e.g., state, county, school) supports partial pooling via random effects. Models are implemented in Stan and fit via rstan'. Provides helpers to validate user-supplied input data, and to predict coverage from fitted models.

Testing differential abundance at individual taxa and in a whole microbial community. The tests are based on the log-ratio of relative abundances. The tests accommodate continuous, discrete (binary, categorical), and multivariate traits, and allow adjustment of confounders. For more details see He (2026) <doi:10.64898/2026.04.07.716976>. This package also includes the com2seq function that combines two sequencing datasets (e.g., from 16S marker-gene and shotgun sequencing) for integrative analysis of differential abundance. For more details see <doi:10.1080/01621459.2025.2516205>.

Support JSON flattening in a long data frame way, where the nesting keys will be stored in the absolute path. It also provides an easy way to summarize the basic description of a JSON list. The idea of mojson is to transform a JSON object in an absolute serialization way, which means the early key-value pairs will appear in the heading rows of the resultant data frame. mojson also provides an alternative way of comparing two different JSON lists, returning the left/inner/right-join style results.

An R interface for the Java Machine Learning for Language Toolkit (mallet) <http://mallet.cs.umass.edu/> to estimate probabilistic topic models, such as Latent Dirichlet Allocation. We can use the R package to read textual data into mallet from R objects, run the Java implementation of mallet directly in R, and extract results as R objects. The Mallet toolkit has many functions, this wrapper focuses on the topic modeling sub-package written by David Mimno. The package uses the rJava package to connect to a JVM.

Power calculations are a critical component of any research study to determine the minimum sample size necessary to detect differences between multiple groups. Here we present an R package, PASSED', that performs power and sample size calculations for the test of two-sample means or ratios with data following beta, gamma (Chang et al. (2011), <doi:10.1007/s00180-010-0209-1>), normal, Poisson (Gu et al. (2008), <doi:10.1002/bimj.200710403>), binomial, geometric, and negative binomial (Zhu and Lakkis (2014), <doi:10.1002/sim.5947>) distributions.

An R interface to pikchr (<https://pikchr.org>, pronounced "picture"), a PIC'-like markup language for creating diagrams within technical documentation. Originally developed by Brian Kernighan, PIC has been adapted into pikchr by D. Richard Hipp, the creator of SQLite'. pikchr is designed to be embedded in fenced code blocks of Markdown or other documentation markup languages, making it ideal for generating diagrams in text-based formats. This package allows R users to seamlessly integrate the descriptive syntax of pikchr for diagram creation directly within the R environment.

Simulate complex data from a given directed acyclic graph and information about each individual node. Root nodes are simply sampled from the specified distribution. Child Nodes are simulated according to one of many implemented regressions, such as logistic regression, linear regression, poisson regression or any other function. Also includes a comprehensive framework for discrete-time simulation, discrete-event simulation, and networks-based simulation which can generate even more complex longitudinal and dependent data. For more details, see Robin Denz, Nina Timmesfeld (2026) <doi:10.18637/jss.v116.i02>.

This package provides a collection of functions for preparing data and fitting Bayesian count spatial regression models, with a specific focus on the Gamma-Count (GC) model. The GC model is well-suited for modeling dispersed count data, including under-dispersed or over-dispersed counts, or counts with equivalent dispersion, using Integrated Nested Laplace Approximations (INLA). The package includes functions for generating data from the GC model, as well as spatially correlated versions of the model. See Nadifar, Baghishani, Fallah (2023) <doi:10.1007/s13253-023-00550-5>.

The framework proposed in Jenul et al., (2022) <doi:10.1007/s10994-022-06221-9>, together with an interactive Shiny dashboard. UBayFS is an ensemble feature selection technique embedded in a Bayesian statistical framework. The method combines data and user knowledge, where the first is extracted via data-driven ensemble feature selection. The user can control the feature selection by assigning prior weights to features and penalizing specific feature combinations. UBayFS can be used for common feature selection as well as block feature selection.

Encodes R values as Concise Binary Object Representation (CBOR, RFC 8949; Bormann and Hoffman (2020) <doi:10.17487/RFC8949>) and decodes CBOR into ordinary R vectors and lists. Decoding checks the whole input against configurable depth, size and item limits before any R object is built, so untrusted input from network peers and devices cannot drive large allocations. Encoding is deterministic: identical R objects always produce identical bytes. Uses a bundled copy of the TinyCBOR library (<https://github.com/intel/tinycbor>), so no system library is required.

BEAST is a Bayesian estimator of abrupt change, seasonality, and trend for decomposing univariate time series and 1D sequential data. Interpretation of time series depends on model choice; different models can yield contrasting or contradicting estimates of patterns, trends, and mechanisms. BEAST alleviates this by abandoning the single-best-model paradigm and instead using Bayesian model averaging over many competing decompositions. It detects and characterizes abrupt changes (changepoints, breakpoints, structural breaks, joinpoints), cyclic or seasonal variation, and nonlinear trends. BEAST not only detects when changes occur but also quantifies how likely the changes are true. It estimates not just piecewise linear trends but also arbitrary nonlinear trends. BEAST is generically applicable to any real-valued time series, such as those from remote sensing, economics, climate science, ecology, hydrology, and other environmental and biological systems. Example applications include identifying regime shifts in ecological data, mapping forest disturbance and land degradation from satellite image time series, detecting market trends in economic indicators, pinpointing anomalies and extreme events in climate records, and analyzing system dynamics in biological time series. Details are given in Zhao et al. (2019) <doi:10.1016/j.rse.2019.04.034>.

Analysis of Ct values from high throughput quantitative real-time PCR (qPCR) assays across multiple conditions or replicates. The input data can be from spatially-defined formats such ABI TaqMan Low Density Arrays or OpenArray; LightCycler from Roche Applied Science; the CFX plates from Bio-Rad Laboratories; conventional 96- or 384-well plates; or microfluidic devices such as the Dynamic Arrays from Fluidigm Corporation. HTqPCR handles data loading, quality assessment, normalization, visualization and parametric or non-parametric testing for statistical significance in Ct values between features (e.g. genes, microRNAs).

LegATo is a suite of open-source software tools for longitudinal microbiome analysis. It is extendable to several different study forms with optimal ease-of-use for researchers. Microbiome time-series data presents distinct challenges including complex covariate dependencies and variety of longitudinal study designs. This toolkit will allow researchers to determine which microbial taxa are affected over time by perturbations such as onset of disease or lifestyle choices, and to predict the effects of these perturbations over time, including changes in composition or stability of commensal bacteria.

The goal of the package aldvmm is to fit adjusted limited dependent variable mixture models of health state utilities. Adjusted limited dependent variable mixture models are finite mixtures of normal distributions with an accumulation of density mass at the limits, and a gap between 100% quality of life and the next smaller utility value. The package aldvmm uses the likelihood and expected value functions proposed by Hernandez Alava and Wailoo (2015) <doi:10.1177/1536867X1501500307> using normal component distributions and a multinomial logit model of probabilities of component membership.

This package provides a Bayesian regression model for discrete response, where the conditional distribution is modelled via a discrete Weibull distribution. This package provides an implementation of Metropolis-Hastings and Reversible-Jumps algorithms to draw samples from the posterior. It covers a wide range of regularizations through any two parameter prior. Examples are Laplace (Lasso), Gaussian (ridge), Uniform, Cauchy and customized priors like a mixture of priors. An extensive visual toolbox is included to check the validity of the results as well as several measures of goodness-of-fit.

Processes multilevel ecological data series (MEDS) arranged in temporal or spatial hierarchies defined by sampling designs. The package recursively applies one-level stage functions to long-format tree-ring and climate data, then optionally fits a terminal model to the processed series. Terminal fits currently include mixed-effects detrending through nlme and Friedman supersmoother detrending through supsmu() of stats'. The package also supports post-detrending comparison and inference tools for multilevel dendroclimatic analyses. The method is described in Lara et al. (2018) <doi:10.1371/journal.pone.0196923>.

This package provides tools for bioinformatics modeling using recursive transformer-inspired architectures, autoencoders, random forests, XGBoost, and stacked ensemble models. Includes utilities for cross-validation, calibration, benchmarking, and threshold optimization in predictive modeling workflows. The methodology builds on ensemble learning (Breiman 2001 <doi:10.1023/A:1010933404324>), gradient boosting (Chen and Guestrin 2016 <doi:10.1145/2939672.2939785>), autoencoders (Hinton and Salakhutdinov 2006 <doi:10.1126/science.1127647>), and recursive transformer efficiency approaches such as Mixture-of-Recursions (Bae et al. 2025 <doi:10.48550/arXiv.2507.10524>).

This package provides an implementation of Congruence Class Models (CCMs) for generating networks. For additional details on CCMs see Goyal, Blitzstein, and De Gruttola (2014) <doi:10.1017/nws.2014.2> and Goyal, De Gruttola, Martin, Rennert, and Onnela <doi:10.48550/arXiv.2603.02467>. ccmnet facilitates sampling networks based on specific topological properties and attribute mixing patterns using a Markov Chain Monte Carlo framework. The implementation builds upon code from the ergm package; see Handcock, Hunter, Butts, Goodreau, and Morris (2008) <doi:10.18637/jss.v024.i01>.

Implementation of some Deep Learning methods. Includes multilayer perceptron, different activation functions, regularisation strategies, stochastic gradient descent and dropout. Thanks go to the following references for helping to inspire and develop the package: Ian Goodfellow, Yoshua Bengio, Aaron Courville, Francis Bach (2016, ISBN:978-0262035613) Deep Learning. Terrence J. Sejnowski (2018, ISBN:978-0262038034) The Deep Learning Revolution. Grant Sanderson (3brown1blue) <https://www.youtube.com/playlist?list=PLZHQObOWTQDNU6R1_67000Dx_ZCJB-3pi> Neural Networks YouTube playlist. Michael A. Nielsen <http://neuralnetworksanddeeplearning.com/> Neural Networks and Deep Learning.

An interactive image editing tool that can be added as part of the HTML in Shiny, R markdown or any type of HTML document. Often times, plots, photos are embedded in the web application/file. drawer can take screenshots of these image-like elements, or any part of the HTML document and send to an image editing space called canvas to allow users immediately edit the screenshot(s) within the same document. Users can quickly combine, compare different screenshots, upload their own images and maybe make a scientific figure.

This package implements a generalized goodness-of-fit test based on spacings for general progressive Type-II censored data. The test statistic is based on the work of Qin et al. (2022) <doi:10.1080/02664763.2020.1821613> and extends the methodology of Balakrishnan et al. (2003) <doi:10.1007/978-1-4612-0103-8_8>. Users can test data against any distribution by providing custom pdf, cdf, and survival functions. The package supports both normal approximation and Monte Carlo simulation approaches for computing p-values and critical values.

Extends classical linear and quadratic discriminant analysis by incorporating permutation-group symmetries into covariance matrix estimation. Methods based on the gips framework identify and impose permutation structures that regularize covariance estimates and improve stability and interpretability for symmetric or exchangeable features. The package provides pooled and class-specific covariance models, including multi-class variants with shared or independently estimated symmetry structures. The underlying methodology is described by Graczyk et al. (2022) <doi:10.1214/22-AOS2174> and Chojecki, Morgen, and KoÅ odziejek (2025) <doi:10.18637/jss.v112.i07>.

An estimation method that can use computer simulations to approximate maximum-likelihood estimates even when the likelihood function can not be evaluated directly. It can be applied whenever it is feasible to conduct many simulations, but works best when the data is approximately Poisson distributed. It was originally designed for demographic inference in evolutionary biology (Naduvilezhath et al., 2011 <doi:10.1111/j.1365-294X.2011.05131.x>, Mathew et al., 2013 <doi:10.1002/ece3.722>). It has optional support for conducting coalescent simulation using the coala package.