PerSIMPER: Per SIMPER function : identification of the main assembly...

View source: R/PerSIMPER.R

PerSIMPERR Documentation

Per SIMPER function : identification of the main assembly process

Description

This function is the basis of DNCImper package and DNCI analysis. It will permute the empirical matrix and produce the empirical as well as the randomized SIMPER profiles. Identify the main assembly process by comparing profiles. Permutations are fixed by rows, columns or both corresponding respectively, to niche, dispersal and niche+dispersal hypothesis. The E index plot is produced to highlight the main assembly process. See Gibert & Escarguel 2019 Global Ecology and Biogeography for theory and more information on process identification. More information in code and comments inside function file.

Usage

PerSIMPER(
  matrixSIMP,
  Groups,
  count = TRUE,
  dataTYPE = "prab",
  Nperm = 1000,
  plotSIMPER = TRUE,
  parallelComputing = FALSE
)

Arguments

matrixSIMP

Sample/Taxa matrix with sample in row and taxa in column

Groups

Grouping vector, ex : c(1,1,1,1,2,2,2,2,2) : 2 groups only !!

count

Display the number of permutation done, can be usefull with very large or small matrix, default = TRUE

dataTYPE

Need to be set for presence/absence or abundance data ("count"), default = "prab" (presence_absence)

Nperm

Number of permutation, default = 1000, should be change to 100 for robustness analysis

plotSIMPER

Display the SIMPER, PerSIMPER and E index plots, default = TRUE

parallelComputing

Run PerSIMPER on half of the available cores/nodes

Examples

A <- DNCImper:::PerSIMPER(Matrix, Group)
#where Matrix is a presence/absence matrix with taxa in column and sample in row
#and Group is a vector with length() == number of rows/samples in Matrix, 2 groups ONLY
#
B <- DNCImper:::PerSIMPER(Matrix, Group, Nperm = 100, count = FALSE, plotSIMPER = FALSE)
#In this example, same data are analysed, with 100 permutations, with no countdown and no plots

Corentin-Gibert-Paleontology/DNCImper documentation built on Feb. 8, 2025, 10:20 a.m.