Using Dirichlet-Multinomial distribution to provide several functions for formal hypothesis testing, power and sample size calculations for human microbiome experiments.
CHANGES IN HMP VERSION 1.6
Added penalty for GA
Changed rpart to use cross validation to find the best tree instead of permutations
Changed KL to use MOM instead of MLE
Fixed getBC so that it drops duplicates instead of crashing
Fixed estPi to handle groups with 1 sample
CHANGES IN HMP VERSION 1.5.1
Added ability to use custom covar distance matrix
Added two new plotting functions
Added text to indicate logging with the pi plot function
Fixed labels in GA plots and results
Fixed bug in Xmcupo.sevsample pvalue calculations
Fixed bug with plot.pi not passing plot title
CHANGES IN HMP VERSION 1.5
Tidied up every function's code, comments, and variable names
Several function parameter names were changed. The old names remain to support backwards compatibility.
kullbackLieber was renamed to Kullback.Leibler. kullbackLieber now just calls Kullback.Leibler.
Added Plot.MDS to plot data
Added Est.PI to estimate the pi vector and plot it with Plot.Pi
Changed data.filter to allow for percentage based filtering
Added a function for testing paired data (Test.Paired)
Added a function that uses rpart to find groups within the data (DM.Rpart and DM.Rpart.Perm)
Added a function that use GA to find taxa that separate 2 groups (DM.GA and DM.GA.Consensus)
DM.MoM can now take a single subject
Added 1 to the numerator and denominators for permutation tests to prevent pvalues = 0
CHANGES IN HMP VERSION 1.4.3
CHANGES IN HMP VERSION 1.4
Changed several functions to improve performance
MC.Xdc.statistic and MC.Xoc.statistics now calculates n.groups themselves
Xmcupo.sevsample now calculates K itself
Added a function to format and combine multiple data sets with different taxa
Added a Kullback Leiber function
CHANGES IN HMP VERSION 1.3.1
CHANGES IN HMP VERSION 1.3
Updated several of the MC functions to perform power calculations on unbalanced designs, however it should be noted that this caused the Nrs parameter to change from a vector to a list for several functions.
Removed the initscalar parameter from the sevsample functions and replaced it with the value of gamma from the DM.MoM function.
Change the names of some functions so they all follow the same formatting. This change mainly affected internal functions.
Changed the wording of the return value in the MC functions from "effect size" to "size of the test statistics" to correctly describe what is being returned.
Changed the Barchart.data function to take a full title rather than just a taxa level.
Changed Xdc.sevsample to select between 'mle' and 'mom' estimators.
Added several warnings/notes clarifying that the value of MC should be >1000 in the MC.functions
Added several warnings/notes clarifying the use of different number of reads in a sample
Added two new functions: Xmcupo.effectsize and pioest