R/intra_samp_phyglm.R

In paternogbc/sensiPhy: Sensitivity Analysis for Comparative Methods

#' Interaction between intraspecific variability and species sampling - Phylogenetic Logistic Regression
#'
#' Performs analyses of sensitivity to species sampling by randomly removing
#' species and detecting the effects on parameter estimates in a phylogenetic
#' logistic regression, while taking into account potential
#' interactions with intraspecific variability.
#'
#' @param formula The model formula
#' @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 n.sim The number of times species are randomly deleted for each
#' \code{break}.
#' @param n.intra Number of datasets resimulated taking into account intraspecific variation (see: \code{"intra_phyglm"}) 
#' @param breaks A vector containing the percentages of species to remove.
#' @param Vx Name of the column containing the standard deviation or the standard error of the predictor 
#' variable. When information is not available for one taxon, the value can be 0 or \code{NA}
#' @param x.transf Transformation for the predictor variable (e.g. \code{"log"} or \code{"sqrt"}). Please use this 
#' argument instead of transforming data in the formula directly (see also details below).
#' @param distrib A character string indicating which distribution to use to generate a random value for the response 
#' and/or predictor variables. Default is normal distribution: "normal" (function \code{\link{rnorm}}).
#' Uniform distribution: "uniform" (\code{\link{runif}})
#' Warning: we recommend to use normal distribution with Vx = standard deviation of the mean.
#' @param btol Bound on searching space. For details see \code{phyloglm}
#' @param track Print a report tracking function progress (default = TRUE)
#' @param ... Further arguments to be passed to \code{phylolm}
#' @details
#'
#' This function randomly removes a given percentage of species (controlled by
#' \code{breaks}) from the full phylogenetic logistic regression, fits a phylogenetic
#' logistic regression model without these species using \code{\link[phylolm]{phylolm}},
#' repeats this many times (controlled by \code{n.sim}), stores the results and
#' calculates the effects on model parameters. 
#' This operation is repeated \code{n.intra} times for simulated values of the dataset, 
#' taking into account intraspecific variation. At each iteration, the function generates a 
#' random value for each row in the dataset using the standard deviation or errors supplied, and 
#' evaluates the effects of sampling within that iteration.
#'
#' All phylogenetic models from \code{phylolm} can be used, i.e. \code{BM},
#' \code{OUfixedRoot}, \code{OUrandomRoot}, \code{lambda}, \code{kappa},
#' \code{delta}, \code{EB} and \code{trend}. See ?\code{phylolm} for details.
#'
#' Currently, this function can only implement simple logistic models (i.e. \eqn{trait~
#' predictor}). In the future we will implement more complex models.
#'
#' Output can be visualised using \code{sensi_plot}.
#' @return The function \code{samp_phyglm} returns a list with the following
#' components:
#' @return \code{formula}: The formula
#' @return \code{full.model.estimates}: Coefficients, aic and the optimised
#' value of the phylogenetic parameter (e.g. \code{lambda} or \code{kappa}) for
#' the full model without deleted species.
#' @return \code{sensi.estimates}: A data frame with all simulation
#' estimates. Each row represents a model rerun with a given number of species
#' \code{n.remov} removed, representing \code{n.percent} of the full dataset.
#' Columns report the calculated regression intercept (\code{intercept}),
#' difference between simulation intercept and full model intercept (\code{DIFintercept}),
#' the percentage of change in intercept compared to the full model (\code{intercept.perc})
#' and intercept p-value (\code{pval.intercept}). All these parameters are also reported
#' for the regression slope (\code{DIFestimate} etc.). Additionally, model aic value
#' (\code{AIC}) and the optimised value (\code{optpar}) of the phylogenetic
#' parameter (e.g. \code{kappa} or \code{lambda}, depending on the phylogenetic model
#' used) are reported. Lastly we reported the standardised difference in intercept 
#' (\code{sDIFintercept}) and slope (\code{sDIFestimate}). 
#' @return \code{sign.analysis} For each break (i.e. each percentage of species
#' removed) this reports the percentage of statistically significant (at p<0.05)
#' intercepts (\code{perc.sign.intercept}) over all repetitions as well as the
#' percentage of statisticaly significant (at p<0.05) slopes (\code{perc.sign.estimate}).
#' @return \code{data}: Original full dataset.
#' @note Please be aware that dropping species may reduce power to detect 
#' significant slopes/intercepts and may partially be responsible for a potential 
#' effect of species removal on p-values. Please also consult standardised differences
#' in the (summary) output.
#' @author Gustavo Paterno, Gijsbert D.A. Werner & Caterina Penone
#' @seealso \code{\link[phylolm]{phylolm}}, \code{\link{samp_phyglm}},
#' \code{\link{intra_phyglm}},\code{\link{intra_samp_phylm}},
#' \code{\link{sensi_plot}}
#' @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
#'
#' Werner, G.D.A., Cornwell, W.K., Sprent, J.I., Kattge, J. & Kiers, E.T. (2014).
#' A single evolutionary innovation drives the deep evolution of symbiotic N2-fixation
#' in angiosperms. Nature Communications, 5, 4087.
#'   
#' Ho, L. S. T. and Ane, C. 2014. "A linear-time algorithm for 
#' Gaussian and non-Gaussian trait evolution models". Systematic Biology 63(3):397-408.
#' 
#' @import ape phylolm
#' 
#' @examples
#' \dontrun{
#' set.seed(6987)
#' phy = rtree(100)
#' x = rTrait(n=1,phy=phy,parameters=list(ancestral.state=2,optimal.value=2,sigma2=1,alpha=1))
#' X = cbind(rep(1,100),x)
#' y = rbinTrait(n=1,phy=phy, beta=c(-1,0.5), alpha=.7 ,X=X)
#' z = rnorm(n = length(x),mean = mean(x),sd = 0.1*mean(x))
#' dat = data.frame(y, x, z)
#' #Run sensitivity analysis:
#' intra_samp <- intra_samp_phyglm(formula = y ~ x, data = dat, phy = phy, 
#'                                n.sim=10, n.intra = 3,
#'                                breaks=seq(.1,.5,.1),
#'                                Vx = "z", distrib="normal", x.transf=NULL)
#' summary(intra_samp)
#' sensi_plot(intra_samp)
#' }

#' @export


intra_samp_phyglm <-
  function(formula,
           data,
           phy,
           n.sim = 10,
           n.intra = 3,
           breaks = seq(.1, .5, .1),
           btol = 50,
           Vx = NULL,
           distrib = "normal",
           x.transf = NULL,
           track = TRUE,
           ...) {
    #Error check
    if (is.null(Vx))
      stop("Vx must be defined")
    if (!inherits(formula, "formula"))
      stop("formula must be class 'formula'")
    if (!inherits(data, "data.frame"))
      stop("data must be class 'data.frame'")
    if (!inherits(phy, "phylo"))
      stop("phy must be class 'phylo'")
    if (formula[[2]] != all.vars(formula)[1] ||
        formula[[3]] != all.vars(formula)[2])
      stop("Please use argument x.transf for data transformation")
    if (distrib == "normal")
      warning ("distrib=normal: make sure that standard deviation is provided for Vx")
    if (length(breaks) < 2)
      stop("Please include more than one break, e.g. breaks=c(.3,.5)")
    
    #Matching tree and phylogeny using utils.R
    datphy <- match_dataphy(formula, data, phy, ...)
    full.data <- datphy[[1]]
    phy <- datphy[[2]]
    
    resp1 <- all.vars(formula)[1]
    if (length(all.vars(formula)) > 2) {
      resp2 <- all.vars(formula)[2]
    }
    pred <- all.vars(formula)[length(all.vars(formula))]
    
    if (!is.null(Vx) && sum(is.na(full.data[, Vx])) != 0) {
      full.data[is.na(full.data[, Vx]), Vx] <- 0
    }
    
    #Function to pick a random value in the interval
    if (distrib == "normal")
      funr <- function(a, b) {
        stats::rnorm(1, a, b)
      }
    else
      funr <- function(a, b) {
        stats::runif(1, a - b, a + b)
      }
    
    
    #List to store information
    intra.samp <- list()
    species.NA <- list()
    
    #Start intra loop here
    errors <- NULL
    if (track == TRUE)
      pb <- utils::txtProgressBar(min = 0, max = n.intra, style = 3)
    counter = 1
    
    for (i in 1:n.intra) {
      ##Set predictor variable
      #Vx is not provided or is not numeric, do not pick random value
      if (!inherits(full.data[, pred], c("numeric", "integer")) ||
          is.null(Vx)) {
        full.data$predV <- full.data[, pred]
      }
      
      #choose a random value in [mean-se,mean+se] if Vx is provided
      if (!is.null(Vx) && is.null(dim(Vx)))
      {
        full.data$predV <-
          apply(full.data[, c(pred, Vx)], 1, function(x)
            funr(x[1], x[2]))
      }
      
      full.data$resp1 <-
        full.data[, resp1] #try to improve this in future
      if (length(all.vars(formula)) > 2) {
        full.data$resp2 <- full.data[, resp2]
      }
      
      #transform Vx if x.transf is provided
      if (!is.null(x.transf))
      {
        suppressWarnings (full.data$predV <- x.transf(full.data$predV))
      }
      
      #skip iteration if there are NA's in the dataset
      species.NA[[i]] <-
        rownames(full.data[with(full.data, is.na(predV)), ])
      if (sum(is.na(full.data[, "predV"]) > 0))
        next
      
      #Run the model
      if (length(all.vars(formula)) > 2) {
        intra.samp[[i]] <-
          samp_phyglm(
            cbind(resp1, resp2) ~ predV,
            data = full.data,
            phy = phy,
            n.sim = n.sim,
            breaks = breaks,
            btol = btol,
            method = "logistic_MPLE",
            track = FALSE,
            verbose = FALSE,
            ...
          )
      } else
        intra.samp[[i]] <-
        samp_phyglm(
          resp1 ~ predV,
          data = full.data,
          phy = phy,
          n.sim = n.sim,
          breaks = breaks,
          btol = btol,
          method = "logistic_MPLE",
          track = FALSE,
          verbose = FALSE,
          ...
        )
      
      if (track == TRUE)
        utils::setTxtProgressBar(pb, counter)
      counter = counter + 1
    }
    
    if (track == TRUE)
      close(pb)
    names(intra.samp) <- 1:n.intra
    
    # Merge lists into data.frames between iterations:
    full.estimates  <-
      suppressWarnings(recombine(intra.samp, slot1 = 4, slot2 = 1))
    influ.estimates <- recombine(intra.samp, slot1 = 5)
    influ.estimates$info <- NULL
    perc.sign <- recombine(intra.samp, slot1 = 6)
    perc.sign$info <- NULL
    
    #Generates output:
    res <- list(
      call = match.call(),
      model = "logistic_MPLE",
      formula = formula,
      full.model.estimates = full.estimates,
      sensi.estimates = influ.estimates,
      sign.analysis = perc.sign,
      data = full.data
    )
    
    
    class(res) <- "sensiIntra_Samp"
    ### Warnings:
    if (length(res$errors) > 0) {
      warning("Some species deletion presented errors, please check: output$errors")
    }
    else {
      res$errors <- "No errors found."
    }
    return(res)
  }