sensiPhy: Sensitivity Analysis for Comparative Methods

Documented in clade_discrete

#' Influential Clade Detection - Trait Evolution Discrete Characters
#' 
#' Fits models for trait evolution of discrete (binary) characters, 
#' detecting influential clades
#'
#' @param data Data frame containing species traits with row names matching tips
#' in \code{phy}.
#' @param phy A phylogeny (class 'phylo') matching \code{data}.
#' @param model The Mkn model to use (see Details). 
#' @param transform The evolutionary model to transform the tree (see Details). Default is \code{none}.
#' @param trait.col The column in the provided data frame which specifies the
#' trait to analyse (which should be a factor with two level)
#' @param clade.col The column in the provided data frame which specifies the
#' clades (a character vector with clade names).
#' @param n.species Minimum number of species in a clade for the clade to be
#' included in the leave-one-out deletion analysis. Default is \code{5}.
#' @param n.sim Number of simulations for the randomization test.
#' @param bounds settings to constrain parameter estimates. See \code{\link[geiger]{fitDiscrete}}
#' @param n.cores number of cores to use. If 'NULL', number of cores is detected.
#' @param track Print a report tracking function progress (default = TRUE)
#' @param ... Further arguments to be passed to \code{\link[geiger]{fitDiscrete}}
#' @details
#' This function sequentially removes one clade at a time,
#' fits a model of discrete character evolution using \code{\link[geiger]{fitDiscrete}}, 
#' repeats this this many times (controlled by \code{n.sim}), stores the results and calculates 
#' the effects on model parameters. Currently, only binary discrete traits are supported. 
#' 
#' Additionally, to account for the influence of the number of species on each 
#' clade (clade sample size), this function also estimates a null distribution
#' expected for the number of species in a given clade. This is done by fitting
#'  models without the same number of species as in the given clade.The number of 
#'  simulations to be performed is set by 'n.sim'. To test if the 
#'  clade influence differs from the null expectation for a clade of that size, 
#'  a randomization test can be performed using 'summary(x)'. 
#'
#' Different character model from \code{fitDiscrete} can be used, including \code{ER} (equal-rates), 
#' \code{SYM} (symmetric), \code{ARD} (all-rates-different) and \code{meristic} (stepwise fashion). 
#'
#' All transformations to the phylogenetic tree from \code{fitDiscrete} can be used, i.e. \code{none},
#' \code{EB}, \code{lambda}, \code{kappa} and\code{delta}.
#' 
#' See \code{\link[geiger]{fitDiscrete}} for more details on character models and tree transformations. 
#' 
#' Output can be visualised using \code{sensi_plot}.
#'
#' @return The function \code{tree_discrete} returns a list with the following
#' components:
#' @return \code{call}: The function call
#' @return \code{data}: The original full data frame. 
#' @return \code{full.model.estimates}: Parameter estimates (transition rates q12 and q21), 
#' AICc and the optimised value of the phylogenetic transformation parameter (e.g. \code{lambda}) 
#' for the full model without deleted clades.
#' @return \code{sensi.estimates}: Parameter estimates (transition rates q12 and q21),(percentual) difference 
#' in parameter estimate compared to the full model (DIFq12, sigsq.q12, DIFq21, optpar.q21), 
#' AICc and the optimised value of the phylogenetic transformation parameter (e.g. \code{lambda}) 
#' for each repeat with a clade removed.
#' @return \code{null.dist}: A data frame with estimates for the null distributions
#' for all clades analysed.
#' @return \code{errors}: Clades where deletion resulted in errors.
#' @return \code{clade.col}: Which column was used to specify the clades?
#' @return \code{optpar}: Transformation parameter used (e.g. \code{lambda}, \code{kappa} etc.)
#' @author Gijsbert Werner & Gustavo Paterno
#' @seealso \code{\link[geiger]{fitDiscrete}}
#' @references 
#' 
#' Paterno, G. B., Penone, C. Werner, G. D. A. 
#' \href{http://doi.wiley.com/10.1111/2041-210X.12990}{sensiPhy: 
#' An r-package for sensitivity analysis in phylogenetic 
#' comparative methods.} Methods in Ecology and Evolution 
#' 2018, 9(6):1461-1467
#' 
#' Yang Z. 2006. Computational Molecular Evolution. Oxford University Press: Oxford. 
#' 
#' Harmon Luke J, Jason T Weir, Chad D Brock, Richard E Glor, and Wendell Challenger. 2008.
#' GEIGER: investigating evolutionary radiations. Bioinformatics 24:129-131.
#' 
#' @examples 
#' \dontrun{
#' #Load data:
#' data("primates")
#' #Create a binary trait factor 
#' primates$data$adultMass_binary<-ifelse(primates$data$adultMass > 7350, "big", "small")
#' clade_disc<-clade_discrete(data=primates$data,phy = primates$phy[[1]],model="SYM",
#' trait.col = "adultMass_binary",clade.col="family",n.sim=30,n.species=10,n.cores = 2)
#' summary(clade_disc)
#' sensi_plot(clade_disc)
#' sensi_plot(clade_disc, clade = "Cebidae", graph = "q12")
#' #Change the evolutionary model, tree transformation or minimum number of species per clade
#' clade_disc_2<-clade_discrete(data=primates$data,phy = primates$phy[[1]],
#' model="ARD",transform="kappa",
#' trait.col = "adultMass_binary",clade.col="family",n.sim=30,
#' n.species=8,n.cores = 2)
#' summary(clade_disc_2)
#' sensi_plot(clade_disc_2)
#' sensi_plot(clade_disc_2, graph = "q12")
#' sensi_plot(clade_disc_2, graph = "q21")
#' }
#' @export


clade_discrete <- function(data,
                           phy,
                           model,
                           transform = "none",
                           trait.col,
                           clade.col,
                           n.species = 5,
                           n.sim = 20,
                           bounds = list(),
                           n.cores = NULL,
                           track = TRUE,
                           ...) {
  # Error checking:
  if (is.null(model))
    stop("model must be specified (e.g. 'ARD' or 'SYM'")
  if (!inherits(data, "data.frame"))
    stop("data must be class 'data.frame'")
  if (missing(clade.col))
    stop("clade.col not defined. Please, define the",
         " column with clade names.")
  if (!inherits(phy, "phylo"))
    stop("phy must be class 'phylo'")
  if (transform == "white")
    stop("the white-noise (non-phylogenetic) model is not allowed")
  if (length(which(!phy$tip.label %in% rownames(data))) > 0)
    stop("not all tips are present in data, prune tree")
  if (length(which(!rownames(data) %in% phy$tip.label)) > 0)
    stop("not all data species are present in tree, remove superfluous data points")
  else
    
    #Calculates the full model, extracts model parameters
    full.data <- data
  phy <- phy
  if (is.na(match(clade.col, names(full.data)))) {
    stop("Names column '", clade.col, "' not found in data frame'")
  }
  
  # Identify CLADES to use and their sample size
  all.clades <- levels(full.data[, clade.col])
  wc <- table(full.data[, clade.col]) > n.species
  uc <- table(full.data[, clade.col])[wc]
  
  #k <- names(which(table(full.data[,clade.col]) > n.species ))
  if (length(uc) == 0)
    stop(
      paste(
        "There is no clade with more than ",
        n.species,
        " species. Change 'n.species' to fix this
                                          problem",
        sep = ""
      )
    )
  
  # FULL MODEL PARAMETERS:
  trait_vec_full <- full.data[, trait.col]
  trait_vec_full <- as.factor(trait_vec_full)
  if (length(levels(trait_vec_full)) > 2)
    stop("discrete data can have maximal two levels")
  names(trait_vec_full) <- rownames(full.data)
  
  N                   <- nrow(full.data)
  mod.0               <-
    geiger::fitDiscrete(
      phy = phy,
      dat = trait_vec_full,
      model = model,
      transform = transform,
      bounds = bounds,
      ncores = n.cores,
      ...
    )
  q12.0               <- mod.0$opt$q12
  q21.0               <- mod.0$opt$q21
  aicc.0              <- mod.0$opt$aicc
  if (transform == "none") {
    optpar.0 <- NA
  }
  if (transform == "EB") {
    optpar.0               <- mod.0$opt$a
  }
  if (transform == "lambda") {
    optpar.0               <- mod.0$opt$lambda
  }
  if (transform == "kappa") {
    optpar.0               <- mod.0$opt$kappa
  }
  if (transform == "delta") {
    optpar.0               <- mod.0$opt$delta
  }
  
  #Create dataframe to store estmates for each clade
  sensi.estimates <-
    data.frame(
      "clade" = I(as.character()),
      "N.species" = numeric(),
      "q12" = numeric(),
      "DIFq12" = numeric(),
      "q12.perc" = numeric(),
      "q21" = numeric(),
      "DIFq21" = numeric(),
      "q21.perc" = numeric(),
      "aicc" = numeric(),
      "optpar" = numeric()
    )
  
  # Create dataframe store simulations (null distribution)
  null.dist <-
    data.frame(
      "clade" = rep(names(uc), each = n.sim),
      "q12" = numeric(length(uc) * n.sim),
      "DIFq12" = numeric(length(uc) * n.sim),
      "q21" = numeric(length(uc) * n.sim),
      "DIFq21" = numeric(length(uc) * n.sim)
    )
  
  
  ### START LOOP between CLADES:
  # counters:
  aa <- 1
  bb <- 1
  errors <- NULL
  
  if (track == TRUE)
    pb <- utils::txtProgressBar(min = 0,
                                max = length(uc) * n.sim,
                                style = 3)
  for (A in names(uc)) {
    ### Number of species in clade A
    cN  <- as.numeric(uc[names(uc) == A])
    
    ### Fit reduced model (without clade)
    crop.data <- full.data[!full.data[, clade.col] %in% A, ]
    crop.phy <-
      ape::drop.tip(phy, setdiff(phy$tip.label, rownames(crop.data)))
    crop.trait_vec <- crop.data[, trait.col]
    crop.trait_vec <- as.factor(crop.trait_vec)
    names(crop.trait_vec) <- rownames(crop.data)
    mod = try(geiger::fitDiscrete(
      phy = crop.phy,
      dat = crop.trait_vec,
      model = model,
      transform = transform,
      bounds = bounds,
      ncores = n.cores,
      ...
    ),
    TRUE)
    q12               <- mod$opt$q12
    q21               <- mod$opt$q21
    DIFq12            <- q12 - q12.0
    DIFq21            <- q21 - q21.0
    q12.perc      <- round((abs(DIFq12 / q12.0)) * 100,
                           digits = 1)
    q21.perc       <- round((abs(DIFq21 / q21.0)) * 100,
                            digits = 1)
    aicc              <- mod$opt$aicc
    if (transform == "none") {
      optpar <- NA
    }
    if (transform == "EB") {
      optpar               <- mod$opt$a
    }
    if (transform == "lambda") {
      optpar               <- mod$opt$lambda
    }
    if (transform == "kappa") {
      optpar               <- mod$opt$kappa
    }
    if (transform == "delta") {
      optpar               <- mod$opt$delta
    }
    
    # Store reduced model parameters:
    estim.simu <- data.frame(A,
                             cN,
                             q12,
                             DIFq12,
                             q12.perc,
                             q21,
                             DIFq21,
                             q21.perc,
                             aicc,
                             optpar,
                             stringsAsFactors = F)
    sensi.estimates[aa,]  <- estim.simu
    
    ### START LOOP FOR NULL DIST:
    # number of species in clade A:
    for (i in 1:n.sim) {
      exclude <- sample(1:N, cN)
      crop.data <- full.data[-exclude, ]
      crop.phy <-
        ape::drop.tip(phy, setdiff(phy$tip.label, rownames(crop.data)))
      crop.trait_vec <- crop.data[, trait.col]
      crop.trait_vec <- as.factor(crop.trait_vec)
      names(crop.trait_vec) <- rownames(crop.data)
      mod = try(geiger::fitDiscrete(
        phy = crop.phy,
        dat = crop.trait_vec,
        model = model,
        transform = transform,
        bounds = bounds,
        ncores = n.cores,
        ...
      ),
      TRUE)
      
      if (isTRUE(class(mod) == "try-error")) {
        error <- i
        names(error) <- rownames(full.data$data)[i]
        errors <- c(errors, error)
        next
      }
      else
        
        q12               <- mod$opt$q12
      q21               <- mod$opt$q21
      aicc              <- mod$opt$aicc
      DIFq12            <- q12 - q12.0
      DIFq21            <- q21 - q21.0
      
      null.dist[bb,] <- data.frame(clade = as.character(A),
                                   q12,
                                   DIFq12,
                                   q21,
                                   DIFq21)
      
      if (track == TRUE)
        utils::setTxtProgressBar(pb, bb)
      bb <- bb + 1
    }
    aa <- aa + 1
  }
  if (track == TRUE)
    on.exit(close(pb))
  
  #OUTPUT
  #full model estimates:
  param0 <- list(
    q12 = q12.0,
    q21 = q21.0,
    aicc = aicc.0,
    optpar = optpar.0
  )
  
  #Generates output:
  res <- list(
    call = match.call(),
    data = full.data,
    full.model.estimates = param0,
    sensi.estimates = sensi.estimates,
    null.dist = null.dist,
    errors = errors,
    optpar = transform,
    clade.col = clade.col
  )
  class(res) <- "sensiClade.TraitEvol"
  ### Warnings:
  if (length(res$errors) > 0) {
    warning("Some clades deletion presented errors, please check: output$errors")
  }
  else {
    res$errors <- "No errors found."
  }
  return(res)
}