RADnormalization_matrix: Normalizes an abundance table to the desired number of ranks
In RADanalysis: Normalization and Study of Rank Abundance Distributions

Description Usage Arguments Value See Also Examples

View source: R/normalization.R

Normalizes an abundance table to the desired number of ranks

1
2
3

RADnormalization_matrix(input, max_rank, average_over = 1, min_rank = 1,
  labels = FALSE, count_data = TRUE, sample_in_row = TRUE,
  method = "upperlimit", verbose = T)

`input`	A vector or matrix which contains the abundance values (an abundance table).
`max_rank`	The desired rank to which this method normalizes the input.
`average_over`	Number of times, a normalized RAD is created and averaged to produce the result.
`min_rank`	The minimum rank to which this method normalizes the input.
`labels`	A logical. If `TRUE` the label of each rank (ids in the input vector) will be returned.
`count_data`	A logical. `TRUE` means that the input vector contains counts (integer values) otherwise contains the relative abundances. In the current version only counts are accepted.
`sample_in_row`	A logical. `TRUE` means that the abundance vector of samples are represented in rows otherwise in columns.
`method`	Sets the stop criterion for normalization. This should be one of "lowerlimit", "middle" or "upperlimit". Method affects the final result. lowerlimit: Samples from species pool one by one, until reaches max_rank. middle: Samples from species pool with random size until the sampled vector has desired ranks (max_rank). upperlimit: Removes from species pool one by one, until reaches max_rank.
`verbose`	A logical. If `TRUE`, prints the progress in percent in console.

A list of following items:

$norm_matrix A matrix which contains normalized RADs sum up to 1. If labels = TRUE, it would also contain the labels.

$inputs A list which contains inputs used for creating normalized RADs. It does not contain input because it could be very big.

RADnormalization for normalize an abundance vector. This function return more details compared to RADnormalization_matrix, representative_point for study the representative of groups of samples in a multi-dimensional scaling plot, representative_RAD for study the representative of group of norm rads.

data("gut_otu_table")
rads <- gut_otu_table
#plot original rads
line_cols <- c("green","red","blue")
sample_classes <- c(1,1,1,1,2,2,3,3,1,1,2,3,3,1,1,2,3,3)
plot(1,xlim = c(1,2000),ylim = c(1,20000),col = "white",log  = "xy",
     xlab = "Rank",ylab = "Abundance",main = "Original RADs from antibiotic data set")
for(i in 1:nrow(rads)){
    temp <- sort(rads[i,],decreasing = TRUE)
    temp <- temp[temp>0]
    lines(x = temp,lwd = 2,col = line_cols[sample_classes[i]])
}
legend("bottomleft",bty = "n",legend = c("pre Cp","under Cp","post Cp"),col = line_cols,lwd = 3)


nrads <- RADnormalization_matrix(input = rads,max_rank = 400,average_over = 20,sample_in_row = TRUE)
nrads <- nrads$norm_matrix

plot(1,xlim = c(1,400),ylim = c(4e-5,1),col = "white",log  = "xy",
     xlab = "Rank",ylab = "Abundance",
     main = "NRADs from antibiotic data set with R = 400 \n with average_over = 20")
for(i in 1:nrow(nrads)){
    lines(x = nrads[i,],lwd = 2,col = line_cols[sample_classes[i]])
}
legend("bottomleft",bty = "n",legend = c("pre Cp","under Cp","post Cp"),col = line_cols,lwd = 3)

1 ( 5.56 %) |2 ( 11.11 %) |3 ( 16.67 %) |4 ( 22.22 %) |5 ( 27.78 %) |6 ( 33.33 %) |7 ( 38.89 %) |8 ( 44.44 %) |9 ( 50 %) |10 ( 55.56 %) |11 ( 61.11 %) |12 ( 66.67 %) |13 ( 72.22 %) |14 ( 77.78 %) |15 ( 83.33 %) |16 ( 88.89 %) |17 ( 94.44 %) |18 ( 100 %) |