ENH: add evenness metrics

.travis.yml

@@ -9,6 +9,8 @@ env:
  global:
    - CRAN: http://cran.rstudio.com
 
+after_failure: "cat /home/travis/build/joey711/phyloseq/phyloseq.Rcheck/00check.log"
+
 notifications:
   email:
     on_success: change

R/extend_vegan.R

@@ -137,7 +137,7 @@ setMethod("vegdist", "phyloseq", function(x, method = "bray", binary = FALSE,
 #'
 #' Performs a number of standard alpha diversity estimates, 
 #' and returns the results as a \code{data.frame}.
-#' Strictly speaking, this function is not only estimating richness,
+#' This function estimates richness and evenness,
 #' despite its name. 
 #' It can operate on the cumulative population of all
 #' samples in the dataset, or by repeating the richness estimates for each
@@ -160,7 +160,7 @@ setMethod("vegdist", "phyloseq", function(x, method = "bray", binary = FALSE,
 #'  Alternatively, you can specify one or more measures
 #'  as a character vector of measure names.
 #'  Values must be among those supported:
-#'  \code{c("Observed", "Chao1", "ACE", "Shannon", "Simpson", "InvSimpson", "Fisher")}.
+#'  \code{c("Observed", "Chao1", "ACE", "Shannon", "Pielou", "Simpson", "InvSimpson", "SimpsonE", "Fisher")}.
 #'
 #' @return A \code{data.frame} of the richness estimates, and their standard error.
 #' 
@@ -175,7 +175,7 @@ setMethod("vegdist", "phyloseq", function(x, method = "bray", binary = FALSE,
 #'  \code{\link[vegan]{diversity}}
 #'  
 #'  \code{\link[vegan]{fisherfit}}
-#'
+#'  
 #' @importFrom vegan estimateR
 #' @importFrom vegan diversity
 #' @importFrom vegan fisher.alpha
@@ -214,8 +214,8 @@ estimate_richness <- function(physeq, split=TRUE, measures=NULL){
 	}
 
 	# Define renaming vector:
-	renamevec = c("Observed", "Chao1", "ACE", "Shannon", "Simpson", "InvSimpson", "Fisher")
-	names(renamevec) <- c("S.obs", "S.chao1", "S.ACE", "shannon", "simpson", "invsimpson", "fisher")
+	renamevec = c("Observed", "Chao1", "ACE", "Shannon", "Pielou", "Simpson", "InvSimpson", "SimpsonE", "Fisher")
+	names(renamevec) <- c("S.obs", "S.chao1", "S.ACE", "shannon", "pielou", "simpson", "invsimpson", "simpsone", "fisher")
 	# If measures was not explicitly provided (is NULL), set to all supported methods
 	if( is.null(measures) ){
 	  measures = as.character(renamevec)
@@ -240,12 +240,20 @@ estimate_richness <- function(physeq, split=TRUE, measures=NULL){
 	if( "Shannon" %in% measures ){
     outlist <- c(outlist, list(shannon = diversity(OTU, index="shannon")))
 	}
+  if( "Pielou" %in% measures){
+  	#print("Starting Pielou")
+  	outlist <- c(outlist, list(pielou = diversity(OTU, index = "shannon")/log(estimateR(OTU)["S.obs",])))
+  }
 	if( "Simpson" %in% measures ){
 	  outlist <- c(outlist, list(simpson = diversity(OTU, index="simpson")))
 	}
 	if( "InvSimpson" %in% measures ){
 	  outlist <- c(outlist, list(invsimpson = diversity(OTU, index="invsimpson")))
 	}
+  if( "SimpsonE" %in% measures ){
+  	#print("Starting SimpsonE")
+  	outlist <- c(outlist, list(simpsone = diversity(OTU, index="invsimpson")/estimateR(OTU)["S.obs",]))
+  }
 	if( "Fisher" %in% measures ){
     fisher = tryCatch(fisher.alpha(OTU, se=TRUE), 
       warning=function(w){

R/plot-methods.R

@@ -557,7 +557,8 @@ plot_net <- function(physeq, distance="bray", type="samples", maxdist = 0.7,
 #'  plot might be kindof strange, and not the intended behavior of this function).
 #'  The following are the names you will want to avoid using in \code{x} or \code{color}:
 #'
-#'  \code{c("Observed", "Chao1", "ACE", "Shannon", "Simpson", "InvSimpson", "Fisher")}.
+#'  \code{c("Observed", "Chao1", "ACE", "Shannon", "Simpson", "Pielou",
+#'  "InvSimpson", "SimpsonE", "Fisher")}.
 #' 
 #' @param physeq (Required). \code{\link{phyloseq-class}}, or alternatively, 
 #'  an \code{\link{otu_table-class}}. The data about which you want to estimate.
@@ -624,7 +625,8 @@ plot_net <- function(physeq, distance="bray", type="samples", maxdist = 0.7,
 #'  Alternatively, you can specify one or more measures
 #'  as a character vector of measure names.
 #'  Values must be among those supported:
-#'  \code{c("Observed", "Chao1", "ACE", "Shannon", "Simpson", "InvSimpson", "Fisher")}.
+#'  \code{c("Observed", "Chao1", "ACE", "Shannon", "Simpson", "Pielou",
+#'  "InvSimpson", "SimpsonE", "Fisher")}.
 #'
 #' @param sortby (Optional). A character string subset of \code{measures} argument.
 #'  Sort x-indices by the mean of one or more \code{measures},

tests/testthat/test-plot.R

@@ -327,7 +327,7 @@ test_that("estimate_richness: test values, classes", {
   expect_is(erdf, "data.frame")
   expect_equivalent(nrow(erdf), 56)
   # Contains all expected measures 
-  expect_true(all(c("Observed", "Chao1", "ACE", "Shannon", "Simpson", "InvSimpson", "Fisher") %in% colnames(erdf)))
+  expect_true(all(c("Observed", "Chao1", "ACE", "Shannon", "Pielou", "Simpson", "InvSimpson", "SimpsonE", "Fisher") %in% colnames(erdf)))
   # and certain standard errors:
   expect_true(all(c("se.chao1", "se.ACE") %in% colnames(erdf)))
   # Test some values. 
@@ -353,7 +353,7 @@ test_that("plot_richness: Standard plots work", {
   p = plot_richness(soilrep)
   expect_is(p, "ggplot")
   expect_equivalent(levels(p$data$variable),
-                    c("Observed", "Chao1", "ACE", "Shannon", "Simpson", "InvSimpson", "Fisher"))
+                    c("Observed", "Chao1", "ACE", "Shannon", "Pielou", "Simpson", "InvSimpson", "SimpsonE", "Fisher"))
   expect_false(all(is.na(p$data$se)))
   expect_true(any(is.na(p$data$se)))
   p = plot_richness(soilrep, measures=c("Observed", "Chao1"))