This function is to change the method to calculate significance between null and observed dissimilarity and/or change the significance threshold values.
change.sigindex(icamp.output, sig.index = c("Confidence", "SES.RC", "SES", "RC"),
detail.save = TRUE, detail.null = FALSE,
ses.cut = 1.96, rc.cut = 0.95, conf.cut = 0.975)list, the exact output of the function icamp.big in which detail.null must be TRUE, to save all null values.
character, Confidence means to directly count the percentage of null values higher/lower than observed value; SES.RC means to use Standard Effect Size (e.g. betaNRI, betaNTI) for phylogenetic beta diversity and use modified Raup-Crick for taxonomic beta diversity, which is typical practice in the previous method; SES means to use Standard Effect Size for both phylogenetic and taxonomic beta diversity; RC means to use modified Raup-Crick for both phylogenetic and taxonomic beta diversity.
logic, whether to output the details, including binning information, significance indexes, bin abundances, and some key parameter settings for iCAMP analysis. Default is TRUE
logic, whether to output all observed and null values of beta diversity indexes. Default is FALSE.
numeric, the cutoff of significant standard effect size, default is 1.96.
numeric, the cutoff of significant modified Raup-Crick index value, default is 0.95.
numeric, the cutoff of significant confidence level (one-tail), default is 0.975.
The output will be the same as icamp.big.
This function is to re-calculate significance using another index or a different threshold value using previously saved null model values. Since the null values are directly extracted from previous icamp.big results, it can skip the most time-consuming step (randomization) and quickly complete calculation.
The default threshold values of Confidence (0.975), SES (1.96), and RC (0.95) mean to capture the 0.95 two-tail confidence level (P=0.05). But, SES need to assume the null values follow normal distribution. RC counts in a half of the special cases that null values are equal to observed values, which is good for obtaining a symmetric metric but theoretically has risk to misestimate significance level (but very slight). Thus, Confidence is preferred as long as the 1000-time randomization is representative.
Ning, D., Yuan, M., Wu, L., Zhang, Y., Guo, X., Zhou, X. et al. (2020). A quantitative framework reveals ecological drivers of grassland microbial community assembly in response to warming. Nature Communications, 11, 4717.
# NOT RUN {
data("icamp.out")
icamp.out.new=change.sigindex(icamp.output=icamp.out, sig.index = "Confidence")
# }
Run the code above in your browser using DataLab