Merge samples by a sample variable or factor

This function provides an alternative to phyloseq::merge_samples() that better handles sample variables of different types, especially categorical sample variables. It combines the samples in x defined by the sample variable or factor group by summing the abundances in otu_table(x) and combines sample variables by the summary functions in funs. The default summary function, unique_or_na(), collapses the values within a group to a single unique value if it exists and otherwise returns NA. The new (merged) samples are named by the values in group.

merge_samples2(x, group, fun_otu = sum, funs = list(), reorder = FALSE)

# S4 method for phyloseq
merge_samples2(x, group, fun_otu = sum, funs = list(), reorder = FALSE)

# S4 method for otu_table
merge_samples2(x, group, fun_otu = sum, reorder = FALSE)

# S4 method for sample_data
merge_samples2(x, group, funs = list(), reorder = FALSE)

Arguments

x	A phyloseq, otu_table, or sample_data object
group	A sample variable or a vector of length nsamples(x) defining the sample grouping. A vector must be supplied if x is an otu_table
fun_otu	Function for combining abundances in the otu table; default is sum. Can be a formula to be converted to a function by purrr::as_mapper()
funs	Named list of merge functions for sample variables; default is unique_or_na
reorder	Logical specifying whether to reorder the new (merged) samples by name

Examples

data(enterotype)

# Merge samples with the same project and clinical status
ps <- enterotype 
sample_data(ps) <- sample_data(ps) %>%
  transform(Project.ClinicalStatus = Project:ClinicalStatus)
sample_data(ps) %>% head
#> Sample Data:        [ 6 samples by 10 sample variables ]:
#>             Enterotype Sample_ID SeqTech SampleID Project Nationality Gender
#> 1 AM.AD.1   NA         AM.AD.1   Sanger  AM.AD.1  gill06  american    F     
#> 2 AM.AD.2   NA         AM.AD.2   Sanger  AM.AD.2  gill06  american    M     
#> 3 AM.F10.T1 NA         AM.F10.T1 Sanger  AM.F10.… turnba… american    F     
#> 4 AM.F10.T2 3          AM.F10.T2 Sanger  AM.F10.… turnba… american    F     
#> 5 DA.AD.1   2          DA.AD.1   Sanger  DA.AD.1  MetaHIT danish      F     
#> 6 DA.AD.1T  NA         DA.AD.1T  Sanger  NA       NA      NA          NA    
#> # … with 3 more variables: Age <dbl>, ClinicalStatus <fct>,
#> #   Project.ClinicalStatus <fct>
#> 
ps0 <- merge_samples2(ps, "Project.ClinicalStatus",
  fun_otu = mean,
  funs = list(Age = mean)
)
#> Error in merge_samples2(ps, "Project.ClinicalStatus", fun_otu = mean,     funs = list(Age = mean)): could not find function "merge_samples2"
sample_data(ps0) %>% head
#> Error in h(simpleError(msg, call)): error in evaluating the argument 'object' in selecting a method for function 'sample_data': object 'ps0' not found