Changes in Version 1.13.1 (2020-08-17)
______________________________________

o When testing for difference in proportion of zeroes, if the Bayesian logistic
  regression fails to converge for a given gene (for example, if there is perfect
  separation between conditions), then NA will be returned (instead of throwing 
  an error. 

Changes in Version 1.3.4 (2018-03-26)
______________________________________

o The test for differences in proportion of zeroes now uses the Wald test 
  p-value instead of the likelihood ratio. This will give very similar results,
  but the Wald test is slightly more conservative.

Changes in Version 1.3.3 (2018-03-23)
______________________________________

o An option has been added to skip the categorization step if only intereseted
  significance of difference. This will speed up computation.
  
o The testing zeroes step and the KS test have been parallelized to speed up
  computation.

Changes in Version 1.1.7 (2017-10-23)
_____________________________________

o A parameter 'level' was added to the main 'scDD' function which allows the 
  user to control the significance level used as the cutoff for considering a 
  gene to be differentially distributed. Previously it was fixed at 0.05 (now 
  the default value).

o Two columns were added to the results object that contain (1) an overall 
  combined p-value (via Fisher's method) for nonzero and zero differences, and 
  (2) Benjamini-Hochberg adjusted version of (1).

o If `testZeroes` is FALSE, the columns with p-values for the test of 
  differential dropout are no longer included. Previously they held NA values.

o If `testZeroes` is TRUE, all zero test p-values are included in the output. 
  Previously, only the tests where the nonzero test was not significant were 
  reported.

Changes in Version 1.1.5 (2017-09-27)
_____________________________________

o The input and output object of the main scDD function has been changed from 
  SummarizedExperiment to SingleCellExperiment to increase interoperability 
  among other Bioconductor packages

o The simulateSet function now returns a SingleCellExperiment object 
  (previously it returned an object in list format)

o The preprocess function now takes as input a SingleCellExperiment object 
  instead of a list of data frames

Changes in Version 1.0.0 (2017-04-23)
_____________________________________

o This is the Bioconductor 3.5 Release version

o scDD is a package for identifying differentially distributed genes in 
  single-cell RNA-seq data. It is designed to detect differences in expression
  that are more complex than a simple mean shift, including:

  - traditional differential expression (DE)

  - differential modality (DM)

  - differential proportion of cells in each state (DP)

  - both differential modality and differential proportion (DB)

Changes in Version 0.99.0 (2016-12-07)
______________________________________

o Initial Bioconductor submission version.