Changes in version 1.25 o Added 'wrenchNorm' function o Added option to use IHW as p-value adustment method in 'MRcoefs' o Modified 'expSummary' slot in 'MRexperiment' object to be of class 'list' instead of 'environment' o Added results classes for 'fitZig' and 'fitFeatureModel' results Changes in version 1.21 o Numerous changes. Added greater flexibility to fitFeatureModel Changes in version 1.15 o Added 'mergeMRexperiment' function o Added 'normFactors' and 'libSize' generics o Added 'fitMultipleTimeSeries' function o Replaced RUnit with testthat library for unit testing o Adding multiple upgrades and changes throughout o Deprecated the load_* functions and created load* function. Changes in version 1.13 o Upgrade support for biom-format vs. 2.0 o Fixed issue - "MRtable, etc will report NA rows when user requests more features than available" o Fixed s2 miscalculation in calcZeroComponent Changes in version 1.11 o Adding fitFeatureModel - a feature based zero-inflated log-normal model. o Added MRcoefs,MRtable,MRfulltable support for fitFeatureModel output. o Added mention in vignette. o Added support for normalizing matrices instead of just MRexperiment objects. o Fixed cumNormStat's non-default qFlag option Changes in version 1.9 o Added flexibility in formula choice for fitTimeSeries o Added readability in ssPermAnalysis o Fixed default in plotClassTimeSeries (include = c("1",...)) o Added fitTimeSeries vignette o Removed interactiveDisplay to namespace - moved to suggests o Fixed ordering of MRtable,MRfulltable first four columns o modified df estimated through responsibilities o renamed fitMeta to fitLogNormal - a more appropriate name Changes in version 1.7 (2014-05-07) o Added function plotBubble o Added parallel (multi-core) options to fitPA, fitDO o Fixed bug for fitMeta when useCSSoffset=FALSE and model matrix ncol==2 o (1.7.10) Updated default quantile estimate (.5) for low estimates o (1.7.10) Added short description on how to do multiple group comparisons o (1.7.15) Output of fitZig (eb) is now a result of limma::eBayes instead of limma::ebayes o (1.7.16) plotMRheatmap allows for sorting by any stat (not just sd) o (1.7.18) fitTimeSeries Including times series method for differentially abundant time intervals o (1.7.20) Fixed minor bug for OTU level time series analyses and added plotClassTimeSeries o (1.7.26) Added warning / fix if any samples are empty in cumNormStat o (1.7.27) Added a few unit tests o (1.7.29) Added interactiveDisplay to namespace (display function allows interactive exploration / plots through browser) Changes in version 1.5 (2014-04-17) o Incorporating biom-format support with the biom2MRexperiment, MRexperiment2biom and load_biome function. o Added uniqueFeatures, filterData, aggregateByTaxonomy / aggTax, plotFeature and calculateEffectiveSamples functions. o Renamed MRfisher to fitPA (presence-absence fisher test). o Added warnings for normalization o Added fitDO (Discovery odds ratio test) and fitMeta (original metastats). o Added match.call() info to fitZig output o Fixed missing E-Step bounds Changes in version 1.2 (2013-08-20) o Our paper got accepted and is available! o Added methods for MRexperiment objects (colSums,colMeans,rowSums,rowMeans, usage is for example colSums(obj) or colSums(obj,norm=TRUE)) (09-25) o Added two new functions, plotOrd and plotRare - a function to plot PCA/MDS coordinates and rarefaction effect (09-04,09-18) o Updated MRfisher to include thresholding for presence-absence testing (08-19) o Updated comments (roxygen2) style for all the functions using the Rd2roxygen package (07-13) o Updated plotCorr and plotMRheatmap to allow various colors/not require trials(07-13) o Rewrote vignette (and switched to knitr) Changes in version 1.1 (2013-06-25) o Rewrote load_meta and load_metaQ to be faster/use less memory o Modified cumNormStat to remove NA samples from calculations (example would be samples without any counts) o Re-added plotGenus' jitter o Fixed uniqueNames call in the MR tables o Changed thanks to Kasper Daniel Hansen's suggestions the following: plotOTU and plotGenus both have much better auto-generated axis MRtable, MRfulltable, MRcoefs have a sort by p-value option now MRtable, MRfulltable, MRcoefs now have an extra option to include unique numbers for OTU features (default would automatically add them previously) cumNorm.R - now returns the object as well - not just replacing the environment 0 Still need to turn the fitZig output to S3, consider subsetting function address low p-values Changes in version 1.0.0 (2013-03-29) o Release!