Changes in version 1.1.3 (2018-10-19) ======================== * Modified COR scores to reflect R^2 for regression on UV/WV/QC rather than max correlation. * Bug fix for shiny report. * Updated vignette. Changes in version 1.1.2 (2017-7-3) ======================== * Moved many Imports: packages to Suggests: to reduce impact of loading scone Changes in version 1.1.1 (2017-6-30) ======================== * Added fast ziber imputation method Changes in version 1.0.0 (2017-4-6) ======================== * Faster SVD with rARPACK. * New zero mode options in scone. * Newly exported get functions. * Updated scone scaling function defaults. * Error handling and documentation updates. * Bug fixes to sample filtering functions. Changes in version 0.99.0 (2016-11-14) ======================== * Major release: many changes due to porting to S4. * Widespread updates to documentation, including examples. * Compatibility with Bioconductor format, including biocViews. * Updated cell-cycle genes. * Default BPPARAM value, now passed as argument to scone. * Removed zinb function. * Test different parallel back ends. * Added class SconeExperiment based on SummarizedExperiment. * Added constructor sconeExperiment to create SconeExperiment objects. * Added many helper methods to retrieve content of slots. * Wrapper get_normalized() to extract/compute single normalization from scone object. * Wrapper get_design() to extract the design matrix associated with a given normalization. * Method select_methods() to get a smaller SconeExperiment object containing only the requested normalization schemes. * biplot_interactive() now works with SconeExperiment objects. * Dropped date from DESCRIPTION as better practice is to add date to NEWS file. * Added wrapper for scran normalization, removed FQP. Changes in version 0.0.8 (2016-09-26) ======================== * Added sconeReport() function for shiny browser of SCONE results. * scone() outputs are now sorted according to mean *rank* of scores rather than mean scores. Although this is a relative measure it accounts for differing variability of metrics. * RLE_IQR now quantifies the variability of the IQR RLE across samples, rather than the mean. * Bug Fix: Previously imputation could be applied to the wrong subset of params rows when user passed params arguments. Imputation functions are now indexed by name to avoid this error. * Bug Fix: Negative infinity expected likelihood is temporarily permitted in the ziber loop. * "ezbake" script and scone_easybake() function added for pipelined SCONE commands. * Revised documentation and removed old scripts. * Various other bug fixes. Changes in version 0.0.6 (2016-07-22) ======================== * Added option for restoring zeroes after scaling step. * New argument format for imputation via impute_args. * Simplified ziber fnr estimation function - requires control genes. * Fixed bug when using plot functionality of filtering functions. * "Conditional" pam replaced with "Stratified" pam, clustering each bio-cross-batch condition separately, rather than simply each bio condition. * Simple FNR for filtering is now based on medians of natural log expression in "expressing" cells / robust to convergence issues. * Removed all sufficient thresholds for metric sample filter. * Added option to write normalized matrices to HDF5 file. * Added wrapper function get_normalized() to retrieve normalized data. * New biplot_interactive function to explore the results. Changes in version 0.0.5 ======================== * Modified biplot to handle general coloring schemes. * Limit number of WV and UV factors to eval_pcs in computing WV and UV scores. * Updated dependencies. * Added error-handling to sample filter. * Removed var preserved measure due to length of running time. Changes in version 0.0.4 ======================== * Fixed a few bugs and documentation mismatches. * Removed stability evaluation (redundant with sil width and slow). * Removed clusterExperiment dependency. * Removed RUV correlation score. UV correlation now takes ruv control genes as default. * Added RLE measures to scone evaluation. * Added FQT_FN to implement careful ties handling by FQ. * Better handling of plots in sample filter functions. * Mean score rather than median rank to evaluate normalizations. * Default value for imputation. * Minor optimizations to evaluation functions. Changes in version 0.0.3 ======================== * Fixed various bugs. * Added “compactness” measure for stability evaluation. * Compute RUV factors only when needed. * Fixed Github issues #11, #12, #13, #14, #21, #28. * zinb now works for non-integer whole numbers. * Updated tests. * Added documentation for datasets. * Added biplot_colored function.