Changes in Version 1.11.6 (2020-07-18) ===================================== * Fixed a bug in the initialization of beta_j. * Fixed a bug in zinbsurf. * Changed zinbwave default to `K=2`. * Fix bug in the initialization of W. Changes in Version 1.7.5 (2019-10-08) ===================================== * Changed default of `zinbwave` to `observationalWeights=FALSE` to speed up computations when weights are not needed. * Added argument `zeroinflation = TRUE`: when set to FALSE a negative binomial model is fit. * Removed dependence on the `copula` package to avoid depending on `gsl`. Changes in Version 1.5.3 (2019-04-19) ===================================== * Updated vignette to reflect new way of interacting with Seurat. * Illustrate the use of approximate `zinbsurf` function. Changes in Version 1.4.2 (2019-03-11) ===================================== * Fix bug in computeObservationalWeights. Changes in Version 1.3.1 (2018-05-09) ===================================== * New `zinbsurf` function implements approximate method for large matrices. * New option `which_genes` in `zinbwave` to specify which genes to use to compute `W`. Changes in Version 1.1.6 (2018-04-17) ===================================== * `zinbwave` now uses `counts` assay by default. * Users can now specify which assay to use to fit the zinb model. Changes in Version 1.1.5 (2018-02-15) ===================================== * Computational weights are computed in `zinbwave` as saved as assay. * Modified vignette to include example of Differential Expression. * Improved documentation for `zinbwave`. Changes in Version 0.99.10 (2017-10-23) ===================================== * Added AIC and BIC to decide number of factors * Added function to compute observational weights for DE Changes in Version 0.99.7 (2017-07-17) ===================================== * `zinbwave()` now returns a `SingleCellExperiment` object. Changes in Version 0.99.6 (2017-07-05) ===================================== * Fixed bug in zinb.loglik.matrix to avoid Inf values Changes in Version 0.99.5 (2017-07-03) ===================================== * Added function computeDevianceResiduals() to compute residuals * Added function imputeZeros() to use the model to impute technical zeros * Added function zinbwave() to perform dimensionality reduction * Changed vignette to illustrate new functions Changes in Version 0.99.4 (2017-06-08) ===================================== * Switch from parallel to BiocParallel * Improvements to code efficiency, e.g., avoid copying ZinbModel objects Changes in Version 0.99.3 (2017-05-31) ===================================== * More informative Description: field * Improved documentation * Added getAlpha, getBeta, and getGamma accessor functions * Improved show() method * Vectorized code in zinbSim() * Fixed bug that introduced an error when initializing an object with empty X or V * Added tests on numerical correctness Changes in Version 0.99.2 (2017-05-12) ===================================== * New formula interface for SummarizedExperiment * Add t-SNE example to vignette Changes in Version 0.99.0 (2017-05-09) ===================================== * Bumped version for submission to Bioconductor * Minor changes to compile vignette * Required R version 3.4 Changes in Version 0.1.4 (2017-04-10) ===================================== * Improved documentation * zinbSim now produces matrix of J x n dimension * Removed unnecessary dependency on clusterExperiment * Additional tests * Better default for epsilon Changes in Version 0.1.2 (2017-03-29) ===================================== * New vignette * Change name to zinbwave Changes in Version 0.1.0 (2016-09-23) ===================================== * Introducing S4 class zinbModel * Major restructuring of the initialization and optimization (see vignette) * Method zinbFit to fit a model * Many other methods, including zinbInitialize, zinbOptimize, zinbSim