NONMEM has been a tool for running nonlinear mixed effects models since the 80s and is still used today (Bauer 2019 <doi:10.1002/psp4.12404>). This tool allows you to convert NONMEM models to rxode2 (Wang, Hallow and James (2016) <doi:10.1002/psp4.12052>) and with simple models nlmixr2 syntax (Fidler et al (2019) <doi:10.1002/psp4.12445>). The nlmixr2 syntax requires the residual specification to be included and it is not always translated. If available, the rxode2 model will read in the NONMEM data and compare the simulation for the population model ('PRED') individual model ('IPRED') and residual model ('IWRES') to immediately show how well the translation is performing. This saves the model development time for people who are creating an rxode2 model manually. Additionally, this package reads in all the information to allow simulation with uncertainty (that is the number of observations, the number of subjects, and the covariance matrix) with a rxode2 model. This is complementary to the babelmixr2 package that translates nlmixr2 models to NONMEM and can convert the objects converted from nonmem2rx to a full nlmixr2 fit.

This package provides tools to simulate and analyse datasets of social interactions between individuals using hierarchical Bayesian models implemented in Stan. The package interacts with Stan via cmdstanr (available from <https://mc-stan.org/r-packages/>) or rstan', depending on user setup. Users can generate realistic interaction data where individual phenotypes influence and respond to those of their partners, with control over sampling design parameters such as the number of individuals, partners, and repeated dyads. The simulation framework allows flexible control over variation and correlation in mean trait values, social responsiveness, and social impact, making it suitable for research on interacting phenotypes and on direct and indirect genetic effects ('DGEs and IGEs'). The package also includes functions to fit and compare alternative models of social effects, including impactâ responsiveness, varianceâ partitioning, and trait-based models, and to summarise model performance in terms of bias and dispersion. For more details on the study of social interactions and impact-responsiveness, see Moore et al. (1997) <doi:10.1111/j.1558-5646.1997.tb01458.x> and de Groot et al. (2022) <doi:10.1016/j.neubiorev.2022.104996>.

clustComp is a package that implements several techniques for the comparison and visualisation of relationships between different clustering results, either flat versus flat or hierarchical versus flat. These relationships among clusters are displayed using a weighted bi-graph, in which the nodes represent the clusters and the edges connect pairs of nodes with non-empty intersection; the weight of each edge is the number of elements in that intersection and is displayed through the edge thickness. The best layout of the bi-graph is provided by the barycentre algorithm, which minimises the weighted number of crossings. In the case of comparing a hierarchical and a non-hierarchical clustering, the dendrogram is pruned at different heights, selected by exploring the tree by depth-first search, starting at the root. Branches are decided to be split according to the value of a scoring function, that can be based either on the aesthetics of the bi-graph or on the mutual information between the hierarchical and the flat clusterings. A mapping between groups of clusters from each side is constructed with a greedy algorithm, and can be additionally visualised.

Allows clustering of incomplete observations by addressing missing values using multiple imputation. For achieving this goal, the methodology consists in three steps, following Audigier and Niang 2022 <doi:10.1007/s11634-022-00519-1>. I) Missing data imputation using dedicated models. Four multiple imputation methods are proposed, two are based on joint modelling and two are fully sequential methods, as discussed in Audigier et al. (2021) <doi:10.48550/arXiv.2106.04424>. II) cluster analysis of imputed data sets. Six clustering methods are available (distances-based or model-based), but custom methods can also be easily used. III) Partition pooling. The set of partitions is aggregated using Non-negative Matrix Factorization based method. An associated instability measure is computed by bootstrap (see Fang, Y. and Wang, J., 2012 <doi:10.1016/j.csda.2011.09.003>). Among applications, this instability measure can be used to choose a number of clusters with missing values. The package also proposes several diagnostic tools to tune the number of imputed data sets, to tune the number of iterations in fully sequential imputation, to check the fit of imputation models, etc.

An implementation of the multilevel (also known as mixed or random effects) hidden Markov model using Bayesian estimation in R. The multilevel hidden Markov model (HMM) is a generalization of the well-known hidden Markov model, for the latter see Rabiner (1989) <doi:10.1109/5.18626>. The multilevel HMM is tailored to accommodate (intense) longitudinal data of multiple individuals simultaneously, see e.g., de Haan-Rietdijk et al. <doi:10.1080/00273171.2017.1370364>. Using a multilevel framework, we allow for heterogeneity in the model parameters (transition probability matrix and conditional distribution), while estimating one overall HMM. The model can be fitted on multivariate data with either a categorical, normal, or Poisson distribution, and include individual level covariates (allowing for e.g., group comparisons on model parameters). Parameters are estimated using Bayesian estimation utilizing the forward-backward recursion within a hybrid Metropolis within Gibbs sampler. Missing data (NA) in the dependent variables is accommodated assuming MAR. The package also includes various visualization options, a function to simulate data, and a function to obtain the most likely hidden state sequence for each individual using the Viterbi algorithm.

Seed germinates through the physical process of water uptake by dry seed driven by the difference in water potential between the seed and the water. There exists seed-to-seed variability in the base seed water potential. Hence, there is a need for a distribution such that a viable seed with its base seed water potential germinates if and only if the soil water potential is more than the base seed water potential. This package estimates the stress tolerance and uniformity parameters of the seed lot for germination under various temperatures by using the hydro-time model of counts of germinated seeds under various water potentials. The distribution of base seed water potential has been considered to follow Normal, Logistic and Extreme value distribution. The estimated proportion of germinated seeds along with the estimates of stress and uniformity parameters are obtained using a generalised linear model. The significance test of the above parameters for within and between temperatures is also performed in the analysis. Details can be found in Kebreab and Murdoch (1999) <doi:10.1093/jxb/50.334.655> and Bradford (2002) <https://www.jstor.org/stable/4046371>.

Sample and cell filtering as well as visualisation of output metrics from Cell Ranger by Grace X.Y. Zheng et al. (2017) <doi:10.1038/ncomms14049>. CRMetrics allows for easy plotting of output metrics across multiple samples as well as comparative plots including statistical assessments of these. CRMetrics allows for easy removal of ambient RNA using SoupX by Matthew D Young and Sam Behjati (2020) <doi:10.1093/gigascience/giaa151> or CellBender by Stephen J Fleming et al. (2022) <doi:10.1101/791699>. Furthermore, it is possible to preprocess data using Pagoda2 by Nikolas Barkas et al. (2021) <https://github.com/kharchenkolab/pagoda2> or Seurat by Yuhan Hao et al. (2021) <doi:10.1016/j.cell.2021.04.048> followed by embedding of cells using Conos by Nikolas Barkas et al. (2019) <doi:10.1038/s41592-019-0466-z>. Finally, doublets can be detected using scrublet by Samuel L. Wolock et al. (2019) <doi:10.1016/j.cels.2018.11.005> or DoubletDetection by Gayoso et al. (2020) <doi:10.5281/zenodo.2678041>. In the end, cells are filtered based on user input for use in downstream applications.

This package provides a variety of original and flexible user-friendly statistical latent variable models for the simultaneous clustering and segmentation of heterogeneous functional data (i.e time series, or more generally longitudinal data, fitted by unsupervised algorithms, including EM algorithms. Functional Latent Data Models for Clustering heterogeneous curves ('FLaMingos') are originally introduced and written in Matlab by Faicel Chamroukhi <https://github.com/fchamroukhi?utf8=?&tab=repositories&q=mix&type=public&language=matlab>. The references are mainly the following ones. Chamroukhi F. (2010) <https://chamroukhi.com/FChamroukhi-PhD.pdf>. Chamroukhi F., Same A., Govaert, G. and Aknin P. (2010) <doi:10.1016/j.neucom.2009.12.023>. Chamroukhi F., Same A., Aknin P. and Govaert G. (2011). <doi:10.1109/IJCNN.2011.6033590>. Same A., Chamroukhi F., Govaert G. and Aknin, P. (2011) <doi:10.1007/s11634-011-0096-5>. Chamroukhi F., and Glotin H. (2012) <doi:10.1109/IJCNN.2012.6252818>. Chamroukhi F., Glotin H. and Same A. (2013) <doi:10.1016/j.neucom.2012.10.030>. Chamroukhi F. (2015) <https://chamroukhi.com/FChamroukhi-HDR.pdf>. Chamroukhi F. and Nguyen H-D. (2019) <doi:10.1002/widm.1298>.

Build spatially and temporally explicit process-based species distribution models, that can include an arbitrary number of environmental factors, species and processes including metabolic constraints and species interactions. The focus of the package is simulating populations of one or multiple species in a grid-based landscape and studying the meta-population dynamics and emergent patterns that arise from the interaction of species under complex environmental conditions. It provides functions for common ecological processes such as negative exponential, kernel-based dispersal (see Nathan et al. (2012) <doi:10.1093/acprof:oso/9780199608898.003.0015>), calculation of the environmental suitability based on cardinal values ( Yin et al. (1995) <doi:10.1016/0168-1923(95)02236-Q>, simplified by Yan and Hunt (1999) <doi:10.1006/anbo.1999.0955> see eq: 4), reproduction in form of an Ricker model (see Ricker (1954) <doi:10.1139/f54-039> and Cabral and Schurr (2010) <doi:10.1111/j.1466-8238.2009.00492.x>), as well as metabolic scaling based on the metabolic theory of ecology (see Brown et al. (2004) <doi:10.1890/03-9000> and Brown, Sibly and Kodric-Brown (2012) <doi:10.1002/9781119968535.ch>).

Identifying outcome relevant subgroups has now become as simple as possible! The formerly lengthy and tedious search for the needle in a haystack will be replaced by a single, comprehensive and coherent presentation. The central result of a subgroup screening is a diagram in which each single dot stands for a subgroup. The diagram may show thousands of them. The position of the dot in the diagram is determined by the sample size of the subgroup and the statistical measure of the treatment effect in that subgroup. The sample size is shown on the horizontal axis while the treatment effect is displayed on the vertical axis. Furthermore, the diagram shows the line of no effect and the overall study results. For small subgroups, which are found on the left side of the plot, larger random deviations from the mean study effect are expected, while for larger subgroups only small deviations from the study mean can be expected to be chance findings. So for a study with no conspicuous subgroup effects, the dots in the figure are expected to form a kind of funnel. Any deviations from this funnel shape hint to conspicuous subgroups.

To optimize clinical trial designs and data analysis methods consistently through trial simulation, we need to simulate multivariate mixed-type virtual patient data independent of designs and analysis methods under evaluation. To make the outcome of optimization more realistic, relevant empirical patient level data should be utilized when itâ s available. However, a few problems arise in simulating trials based on small empirical data, where the underlying marginal distributions and their dependence structure cannot be understood or verified thoroughly due to the limited sample size. To resolve this issue, we use the copula invariance property, which can generate the joint distribution without making a strong parametric assumption. The function copula.sim can generate virtual patient data with optional data validation methods that are based on energy distance and ball divergence measurement. The function compare.copula.sim can conduct comparison of marginal mean and covariance of simulated data. To simulate patient-level data from a hypothetical treatment arm that would perform differently from the observed data, the function new.arm.copula.sim can be used to generate new multivariate data with the same dependence structure of the original data but with a shifted mean vector.

This package provides users a quick exploratory dive into common visualizations without writing a single line of code given the users data follows the Analysis Data Model (ADaM) standards put forth by the Clinical Data Interchange Standards Consortium (CDISC) <https://www.cdisc.org>. Prominent modules/ features of the application are the Table Generator, Population Explorer, and the Individual Explorer. The Table Generator allows users to drag and drop variables and desired statistics (frequencies, means, ANOVA, t-test, and other summary statistics) into bins that automagically create stunning tables with validated information. The Population Explorer offers various plots to visualize general trends in the population from various vantage points. Plot modules currently include scatter plot, spaghetti plot, box plot, histogram, means plot, and bar plot. Each plot type allows the user to plot uploaded variables against one another, and dissect the population by filtering out certain subjects. Last, the Individual Explorer establishes a cohesive patient narrative, allowing the user to interact with patient metrics (params) by visit or plotting important patient events on a timeline. All modules allow for concise filtering & downloading bulk outputs into html or pdf formats to save for later.

This package provides tools for transforming, a posteriori time-scaling, and modifying phylogenies containing extinct (i.e. fossil) lineages. In particular, most users are interested in the functions timePaleoPhy, bin_timePaleoPhy, cal3TimePaleoPhy and bin_cal3TimePaleoPhy, which date cladograms of fossil taxa using stratigraphic data. This package also contains a large number of likelihood functions for estimating sampling and diversification rates from different types of data available from the fossil record (e.g. range data, occurrence data, etc). paleotree users can also simulate diversification and sampling in the fossil record using the function simFossilRecord, which is a detailed simulator for branching birth-death-sampling processes composed of discrete taxonomic units arranged in ancestor-descendant relationships. Users can use simFossilRecord to simulate diversification in incompletely sampled fossil records, under various models of morphological differentiation (i.e. the various patterns by which morphotaxa originate from one another), and with time-dependent, longevity-dependent and/or diversity-dependent rates of diversification, extinction and sampling. Additional functions allow users to translate simulated ancestor-descendant data from simFossilRecord into standard time-scaled phylogenies or unscaled cladograms that reflect the relationships among taxon units.

We quantitatively evaluated the assertion that says if one suit is found to be evenly distributed among the 4 players, the rest of the suits are more likely to be evenly distributed. Our mathematical analyses show that, if one suit is found to be evenly distributed, then a second suit has a slightly elevated probability (ranging between 10% to 15%) of being evenly distributed. If two suits are found to be evenly distributed, then a third suit has a substantially elevated probability (ranging between 30% to 50%) of being evenly distributed.This package refers to methods and authentic data from Ely Culbertson <https://www.bridgebum.com/law_of_symmetry.php>, Gregory Stoll <https://gregstoll.com/~gregstoll/bridge/math.html>, and details of performing the probability calculations from Jeremy L. Martin <https://jlmartin.ku.edu/~jlmartin/bridge/basics.pdf>, Emile Borel and Andre Cheron (1954) "The Mathematical Theory of Bridge",Antonio Vivaldi and Gianni Barracho (2001, ISBN:0 7134 8663 5) "Probabilities and Alternatives in Bridge", Ken Monzingo (2005) "Hand and Suit Patterns" <http://web2.acbl.org/documentlibrary/teachers/celebritylessons/handpatternsrevised.pdf>Ken Monzingo (2005) "Hand and Suit Patterns" <http://web2.acbl.org/documentlibrary/teachers/celebritylessons/handpatternsrevised.pdf>.

The git2rdata package is an R package for writing and reading dataframes as plain text files. A metadata file stores important information. 1) Storing metadata allows to maintain the classes of variables. By default, git2rdata optimizes the data for file storage. The optimization is most effective on data containing factors. The optimization makes the data less human readable. The user can turn this off when they prefer a human readable format over smaller files. Details on the implementation are available in vignette("plain_text", package = "git2rdata"). 2) Storing metadata also allows smaller row based diffs between two consecutive commits. This is a useful feature when storing data as plain text files under version control. Details on this part of the implementation are available in vignette("version_control", package = "git2rdata"). Although we envisioned git2rdata with a git workflow in mind, you can use it in combination with other version control systems like subversion or mercurial. 3) git2rdata is a useful tool in a reproducible and traceable workflow. vignette("workflow", package = "git2rdata") gives a toy example. 4) vignette("efficiency", package = "git2rdata") provides some insight into the efficiency of file storage, git repository size and speed for writing and reading.

Density-based clustering methods are well adapted to the clustering of high-dimensional data and enable the discovery of core groups of various shapes despite large amounts of noise. This package provides a novel density-based cluster extraction method, OPTICS k-Xi, and a framework to compare k-Xi models using distance-based metrics to investigate datasets with unknown number of clusters. The vignette first introduces density-based algorithms with simulated datasets, then presents and evaluates the k-Xi cluster extraction method. Finally, the models comparison framework is described and experimented on 2 genetic datasets to identify groups and their discriminating features. The k-Xi algorithm is a novel OPTICS cluster extraction method that specifies directly the number of clusters and does not require fine-tuning of the steepness parameter as the OPTICS Xi method. Combined with a framework that compares models with varying parameters, the OPTICS k-Xi method can identify groups in noisy datasets with unknown number of clusters. Results on summarized genetic data of 1,200 patients are in Charlon T. (2019) <doi:10.13097/archive-ouverte/unige:161795>. A short video tutorial can be found at <https://www.youtube.com/watch?v=P2XAjqI5Lc4/>.

Generates predictive distributions based on calibrating priors for various commonly used statistical models, including models with predictors. Routines for densities, probabilities, quantiles, random deviates and the parameter posterior are provided. The predictions are generated from the Bayesian prediction integral, with priors chosen to give good reliability (also known as calibration). For homogeneous models, the prior is set to the right Haar prior, giving predictions which are exactly reliable. As a result, in repeated testing, the frequencies of out-of-sample outcomes and the probabilities from the predictions agree. For other models, the prior is chosen to give good reliability. Where possible, the Bayesian prediction integral is solved exactly. Where exact solutions are not possible, the Bayesian prediction integral is solved using the Datta-Mukerjee-Ghosh-Sweeting (DMGS) asymptotic expansion. Optionally, the prediction integral can also be solved using posterior samples generated using Paul Northrop's ratio of uniforms sampling package ('rust'). Results are also generated based on maximum likelihood, for comparison purposes. Various model selection diagnostics and testing routines are included. Based on "Reducing reliability bias in assessments of extreme weather risk using calibrating priors", Jewson, S., Sweeting, T. and Jewson, L. (2024); <doi:10.5194/ascmo-11-1-2025>.

This package provides a powerful yet simple graphical tool available in the field of psychometrics is the Wright Map (also known as item maps or item-person maps), which presents the location of both respondents and items on the same scale. Wright Maps are commonly used to present the results of dichotomous or polytomous item response models. The WrightMap package provides functions to create these plots from item parameters and person estimates stored as R objects. Although the package can be used in conjunction with any software used to estimate the IRT model (e.g. TAM', mirt', eRm or IRToys in R', or Stata', Mplus', etc.), WrightMap features special integration with ConQuest to facilitate reading and plotting its output directly.The wrightMap function creates Wright Maps based on person estimates and item parameters produced by an item response analysis. The CQmodel function reads output files created using ConQuest software and creates a set of data frames for easy data manipulation, bundled in a CQmodel object. The wrightMap function can take a CQmodel object as input or it can be used to create Wright Maps directly from data frames of person and item parameters.

Contain code to work with a C struct, in short cgeneric, to define a Gaussian Markov random (GMRF) model. The cgeneric contain code to specify GMRF elements such as the graph and the precision matrix, and also the initial and prior for its parameters, useful for model inference. It can be accessed from a C program and is the recommended way to implement new GMRF models in the INLA package (<https://www.r-inla.org>). The INLAtools implement functions to evaluate each one of the model specifications from R. The implemented functionalities leverage the use of cgeneric models and provide a way to debug the code as well to work with the prior for the model parameters and to sample from it. A very useful functionality is the Kronecker product method that creates a new model from multiple cgeneric models. It also works with the rgeneric, the R version of the cgeneric intended to easy try implementation of new GMRF models. The Kronecker between two cgeneric models was used in Sterrantino et. al. (2024) <doi:10.1007/s10260-025-00788-y>, and can be used to build the spatio-temporal intrinsic interaction models for what the needed constraints are automatically set.

The identification of novel compound-protein interaction (CPI) is important in drug discovery. Revealing unknown compound-protein interactions is useful to design a new drug for a target protein by screening candidate compounds. The accurate CPI prediction assists in effective drug discovery process. To identify potential CPI effectively, prediction methods based on machine learning and deep learning have been developed. Data for sequences are provided as discrete symbolic data. In the data, compounds are represented as SMILES (simplified molecular-input line-entry system) strings and proteins are sequences in which the characters are amino acids. The outcome is defined as a variable that indicates how strong two molecules interact with each other or whether there is an interaction between them. In this package, a deep-learning based model that takes only sequence information of both compounds and proteins as input and the outcome as output is used to predict CPI. The model is implemented by using compound and protein encoders with useful features. The CPI model also supports other modeling tasks, including protein-protein interaction (PPI), chemical-chemical interaction (CCI), or single compounds and proteins. Although the model is designed for proteins, DNA and RNA can be used if they are represented as sequences.

This package provides functions for the import, transformation, and analysis of data from muscle physiology experiments. The work loop technique is used to evaluate the mechanical work and power output of muscle. Josephson (1985) <doi:10.1242/jeb.114.1.493> modernized the technique for application in comparative biomechanics. Although our initial motivation was to provide functions to analyze work loop experiment data, as we developed the package we incorporated the ability to analyze data from experiments that are often complementary to work loops. There are currently three supported experiment types: work loops, simple twitches, and tetanus trials. Data can be imported directly from .ddf files or via an object constructor function. Through either method, data can then be cleaned or transformed via methods typically used in studies of muscle physiology. Data can then be analyzed to determine the timing and magnitude of force development and relaxation (for isometric trials) or the magnitude of work, net power, and instantaneous power among other things (for work loops). Although we do not provide plotting functions, all resultant objects are designed to be friendly to visualization via either base-R plotting or tidyverse functions. This package has been peer-reviewed by rOpenSci (v. 1.1.0).

For Bayesian and classical inference and prediction with count-valued data, Simultaneous Transformation and Rounding (STAR) Models provide a flexible, interpretable, and easy-to-use approach. STAR models the observed count data using a rounded continuous data model and incorporates a transformation for greater flexibility. Implicitly, STAR formalizes the commonly-applied yet incoherent procedure of (i) transforming count-valued data and subsequently (ii) modeling the transformed data using Gaussian models. STAR is well-defined for count-valued data, which is reflected in predictive accuracy, and is designed to account for zero-inflation, bounded or censored data, and over- or underdispersion. Importantly, STAR is easy to combine with existing MCMC or point estimation methods for continuous data, which allows seamless adaptation of continuous data models (such as linear regressions, additive models, BART, random forests, and gradient boosting machines) for count-valued data. The package also includes several methods for modeling count time series data, namely via warped Dynamic Linear Models. For more details and background on these methodologies, see the works of Kowal and Canale (2020) <doi:10.1214/20-EJS1707>, Kowal and Wu (2022) <doi:10.1111/biom.13617>, King and Kowal (2022) <arXiv:2110.14790>, and Kowal and Wu (2023) <arXiv:2110.12316>.

Read CODAR's SeaSonde High-Frequency Radar spectra files, compute radial metrics, and generate plots for spectra and antenna pattern data. Implementation is based in technical manuals, publications and patents, please refer to the following documents for more information: Barrick and Lipa (1999) <https://codar.com/images/about/patents/05990834.PDF>; CODAR Ocean Sensors (2002) <http://support.codar.com/Technicians_Information_Page_for_SeaSondes/Docs/Informative/FirstOrder_Settings.pdf>; Lipa et al. (2006) <doi:10.1109/joe.2006.886104>; Paolo et al. (2007) <doi:10.1109/oceans.2007.4449265>; CODAR Ocean Sensors (2009a) <http://support.codar.com/Technicians_Information_Page_for_SeaSondes/Docs/GuidesToFileFormats/File_AntennaPattern.pdf>; CODAR Ocean Sensors (2009b) <http://support.codar.com/Technicians_Information_Page_for_SeaSondes/Docs/GuidesToFileFormats/File_CrossSpectraReduced.pdf>; CODAR Ocean Sensors (2016a) <http://support.codar.com/Technicians_Information_Page_for_SeaSondes/Manuals_Documentation_Release_8/File_Formats/File_Cross_Spectra_V6.pdf>; CODAR Ocean Sensors (2016b) <http://support.codar.com/Technicians_Information_Page_for_SeaSondes/Manuals_Documentation_Release_8/File_Formats/FIle_Reduced_Spectra.pdf>; CODAR Ocean Sensors (2016c) <http://support.codar.com/Technicians_Information_Page_for_SeaSondes/Manuals_Documentation_Release_8/Application_Guides/Guide_SpectraPlotterMap.pdf>; Bushnell and Worthington (2022) <doi:10.25923/4c5x-g538>.

Test for association between the observed data and their estimated latent variables. The jackstraw package provides a resampling strategy and testing scheme to estimate statistical significance of association between the observed data and their latent variables. Depending on the data type and the analysis aim, the latent variables may be estimated by principal component analysis (PCA), factor analysis (FA), K-means clustering, and related unsupervised learning algorithms. The jackstraw methods learn over-fitting characteristics inherent in this circular analysis, where the observed data are used to estimate the latent variables and used again to test against that estimated latent variables. When latent variables are estimated by PCA, the jackstraw enables statistical testing for association between observed variables and latent variables, as estimated by low-dimensional principal components (PCs). This essentially leads to identifying variables that are significantly associated with PCs. Similarly, unsupervised clustering, such as K-means clustering, partition around medoids (PAM), and others, finds coherent groups in high-dimensional data. The jackstraw estimates statistical significance of cluster membership, by testing association between data and cluster centers. Clustering membership can be improved by using the resulting jackstraw p-values and posterior inclusion probabilities (PIPs), with an application to unsupervised evaluation of cell identities in single cell RNA-seq (scRNA-seq).