Get appropriate numeric transformation function
get_numeric_transform(transform, ...)character string or function, or list
that may contain character string or function.
additional arguments are ignored.
function or NULL when no matching function is
found, or list is returned when the input transform
has multiple values.
This function recognizes numeric transformation functions by name, or searches the attached R package environments for a matching function by name.
Recognized transform names:
"none" or "linear" returns the data without change
"log2signed" applies jamba::log2signed(), defined as
log2(1+x) transform to the absolute value, then multiplies
by the original sign(x)
"exp2signed" applies the inverse of "log2signed", which
exponentiates numeric values which were previously transformed
with log2(1+x). Note that value=0 when exponentiated becomes +1
"sqrt" applies square root transform, equivalent to sqrt()
"cubert" applies cube root x^(1/3)
"qrt" applies fourth root x^(1/4)
"frt" or "fthrt" applies fifth root x^(1/5)
"square" applies x^2 to absolute value, multiplied by the sign(x)
"cube" applies x^3;
Any other character name is used to find a function with the same
name, and if the function is found it is used. The function is
expected to take input x and return corresponding output with the
same length as the input. A function such as max() does not fit
this criteria, but a function such as log2() is acceptable.
Other jam utility functions:
cardinality(),
color_complement(),
convert_PD_df_to_SE(),
convert_imputed_assays_to_na(),
curate_se_colData(),
curate_to_df_by_pattern(),
design2layout(),
handle_df_args(),
merge_proteomics_se(),
rowNormScale()
get_numeric_transform("log2signed")
#> function
#> (x,
#> offset=1,
#> base=2,
#> ...)
#> {
#> ## Purpose is to transform numeric data using log2 transformation
#> ## but where negative values are kept negative by log2-transforming
#> ## the absolute value, then multiplying by the original sign.
#> if (length(x) == 0) {
#> return(x);
#> }
#> if (offset < 1 && any(abs(x) < 1)) {
#> stop(
#> paste0(
#> "Values in abs(x) less than offset ",
#> offset,
#> " cannot be transformed without losing direction.")
#> );
#> }
#>
#> ## Determine the sign(x)
#> x_sign <- sign(x);
#> ## For now, do not convert sign 0 to sign 1.
#> #x_sign <- ifelse(x_sign == 0, 1, x_sign);
#>
#> if (length(base) == 0 || all(unique(base) == 2)) {
#> return(log2(abs(x) + offset) * x_sign);
#> }
#> # Note: the conversion to different log base
#> #log2(abs(x) + offset) * x_sign / log2(base);
#> log(abs(x) + offset, base=base) * x_sign;
#> }
#> <bytecode: 0x1719b5710>
#> <environment: namespace:jamba>
transform_list <- get_numeric_transform(
c("none",
"log2signed",
log2,
sqrt));
jamba::sdim(transform_list);
#> rows class
#> none 1 function
#> log2signed 1 function
#> log2 1 function
#> sqrt 1 function
x <- jamba::nameVector(0:16);
transform_list[[1]](x)
#> 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16
#> 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16
transform_list[[2]](x)
#> 0 1 2 3 4 5 6 7
#> 0.000000 1.000000 1.584963 2.000000 2.321928 2.584963 2.807355 3.000000
#> 8 9 10 11 12 13 14 15
#> 3.169925 3.321928 3.459432 3.584963 3.700440 3.807355 3.906891 4.000000
#> 16
#> 4.087463