Mercurial > repos > eschen42 > w4mkmeans

annotate w4m_general_purpose_routines.R @ 2:c415b7dc6f37 draft default tip

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
planemo upload for repository https://github.com/HegemanLab/w4mkmeans_galaxy_wrapper/tree/master commit 3e916537da6bb37e6f3927d7a11e98e0ab6ef5ec
author eschen42
date Mon, 05 Mar 2018 12:40:17 -0500
parents 6ccbe18131a6
children
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
2
c415b7dc6f37 planemo upload for repository https://github.com/HegemanLab/w4mkmeans_galaxy_wrapper/tree/master commit 3e916537da6bb37e6f3927d7a11e98e0ab6ef5ec
eschen42
parents: 0
diff changeset
1 ##-----------------------------------------------
c415b7dc6f37 planemo upload for repository https://github.com/HegemanLab/w4mkmeans_galaxy_wrapper/tree/master commit 3e916537da6bb37e6f3927d7a11e98e0ab6ef5ec
eschen42
parents: 0
diff changeset
2 ## helper functions for error detection/reporting
c415b7dc6f37 planemo upload for repository https://github.com/HegemanLab/w4mkmeans_galaxy_wrapper/tree/master commit 3e916537da6bb37e6f3927d7a11e98e0ab6ef5ec
eschen42
parents: 0
diff changeset
3 ##-----------------------------------------------
c415b7dc6f37 planemo upload for repository https://github.com/HegemanLab/w4mkmeans_galaxy_wrapper/tree/master commit 3e916537da6bb37e6f3927d7a11e98e0ab6ef5ec
eschen42
parents: 0
diff changeset
4
c415b7dc6f37 planemo upload for repository https://github.com/HegemanLab/w4mkmeans_galaxy_wrapper/tree/master commit 3e916537da6bb37e6f3927d7a11e98e0ab6ef5ec
eschen42
parents: 0
diff changeset
5 # ISO 8601 date ref: https://en.wikipedia.org/wiki/ISO_8601
c415b7dc6f37 planemo upload for repository https://github.com/HegemanLab/w4mkmeans_galaxy_wrapper/tree/master commit 3e916537da6bb37e6f3927d7a11e98e0ab6ef5ec
eschen42
parents: 0
diff changeset
6 iso_date <- function() {
c415b7dc6f37 planemo upload for repository https://github.com/HegemanLab/w4mkmeans_galaxy_wrapper/tree/master commit 3e916537da6bb37e6f3927d7a11e98e0ab6ef5ec
eschen42
parents: 0
diff changeset
7 format(Sys.time(), "%Y-%m-%dT%H:%M:%S%z")
c415b7dc6f37 planemo upload for repository https://github.com/HegemanLab/w4mkmeans_galaxy_wrapper/tree/master commit 3e916537da6bb37e6f3927d7a11e98e0ab6ef5ec
eschen42
parents: 0
diff changeset
8 }
c415b7dc6f37 planemo upload for repository https://github.com/HegemanLab/w4mkmeans_galaxy_wrapper/tree/master commit 3e916537da6bb37e6f3927d7a11e98e0ab6ef5ec
eschen42
parents: 0
diff changeset
9
c415b7dc6f37 planemo upload for repository https://github.com/HegemanLab/w4mkmeans_galaxy_wrapper/tree/master commit 3e916537da6bb37e6f3927d7a11e98e0ab6ef5ec
eschen42
parents: 0
diff changeset
10 # log-printing to stderr
c415b7dc6f37 planemo upload for repository https://github.com/HegemanLab/w4mkmeans_galaxy_wrapper/tree/master commit 3e916537da6bb37e6f3927d7a11e98e0ab6ef5ec
eschen42
parents: 0
diff changeset
11 log_print <- function(x, ...) {
c415b7dc6f37 planemo upload for repository https://github.com/HegemanLab/w4mkmeans_galaxy_wrapper/tree/master commit 3e916537da6bb37e6f3927d7a11e98e0ab6ef5ec
eschen42
parents: 0
diff changeset
12 cat(
c415b7dc6f37 planemo upload for repository https://github.com/HegemanLab/w4mkmeans_galaxy_wrapper/tree/master commit 3e916537da6bb37e6f3927d7a11e98e0ab6ef5ec
eschen42
parents: 0
diff changeset
13 sep=""
c415b7dc6f37 planemo upload for repository https://github.com/HegemanLab/w4mkmeans_galaxy_wrapper/tree/master commit 3e916537da6bb37e6f3927d7a11e98e0ab6ef5ec
eschen42
parents: 0
diff changeset
14 , file=stderr()
c415b7dc6f37 planemo upload for repository https://github.com/HegemanLab/w4mkmeans_galaxy_wrapper/tree/master commit 3e916537da6bb37e6f3927d7a11e98e0ab6ef5ec
eschen42
parents: 0
diff changeset
15 , iso_date()
c415b7dc6f37 planemo upload for repository https://github.com/HegemanLab/w4mkmeans_galaxy_wrapper/tree/master commit 3e916537da6bb37e6f3927d7a11e98e0ab6ef5ec
eschen42
parents: 0
diff changeset
16 , " "
c415b7dc6f37 planemo upload for repository https://github.com/HegemanLab/w4mkmeans_galaxy_wrapper/tree/master commit 3e916537da6bb37e6f3927d7a11e98e0ab6ef5ec
eschen42
parents: 0
diff changeset
17 , c(x, ...)
c415b7dc6f37 planemo upload for repository https://github.com/HegemanLab/w4mkmeans_galaxy_wrapper/tree/master commit 3e916537da6bb37e6f3927d7a11e98e0ab6ef5ec
eschen42
parents: 0
diff changeset
18 , "\n"
c415b7dc6f37 planemo upload for repository https://github.com/HegemanLab/w4mkmeans_galaxy_wrapper/tree/master commit 3e916537da6bb37e6f3927d7a11e98e0ab6ef5ec
eschen42
parents: 0
diff changeset
19 )
c415b7dc6f37 planemo upload for repository https://github.com/HegemanLab/w4mkmeans_galaxy_wrapper/tree/master commit 3e916537da6bb37e6f3927d7a11e98e0ab6ef5ec
eschen42
parents: 0
diff changeset
20 }
c415b7dc6f37 planemo upload for repository https://github.com/HegemanLab/w4mkmeans_galaxy_wrapper/tree/master commit 3e916537da6bb37e6f3927d7a11e98e0ab6ef5ec
eschen42
parents: 0
diff changeset
21
c415b7dc6f37 planemo upload for repository https://github.com/HegemanLab/w4mkmeans_galaxy_wrapper/tree/master commit 3e916537da6bb37e6f3927d7a11e98e0ab6ef5ec
eschen42
parents: 0
diff changeset
22 # format error for logging
c415b7dc6f37 planemo upload for repository https://github.com/HegemanLab/w4mkmeans_galaxy_wrapper/tree/master commit 3e916537da6bb37e6f3927d7a11e98e0ab6ef5ec
eschen42
parents: 0
diff changeset
23 format_error <- function(e) {
c415b7dc6f37 planemo upload for repository https://github.com/HegemanLab/w4mkmeans_galaxy_wrapper/tree/master commit 3e916537da6bb37e6f3927d7a11e98e0ab6ef5ec
eschen42
parents: 0
diff changeset
24 paste(c("Error { message:", e$message, ", call:", e$call, "}"), collapse = " ")
c415b7dc6f37 planemo upload for repository https://github.com/HegemanLab/w4mkmeans_galaxy_wrapper/tree/master commit 3e916537da6bb37e6f3927d7a11e98e0ab6ef5ec
eschen42
parents: 0
diff changeset
25 }
c415b7dc6f37 planemo upload for repository https://github.com/HegemanLab/w4mkmeans_galaxy_wrapper/tree/master commit 3e916537da6bb37e6f3927d7a11e98e0ab6ef5ec
eschen42
parents: 0
diff changeset
26
c415b7dc6f37 planemo upload for repository https://github.com/HegemanLab/w4mkmeans_galaxy_wrapper/tree/master commit 3e916537da6bb37e6f3927d7a11e98e0ab6ef5ec
eschen42
parents: 0
diff changeset
27 # tryCatchFunc produces a list
c415b7dc6f37 planemo upload for repository https://github.com/HegemanLab/w4mkmeans_galaxy_wrapper/tree/master commit 3e916537da6bb37e6f3927d7a11e98e0ab6ef5ec
eschen42
parents: 0
diff changeset
28 # func - a function that takes no arguments
c415b7dc6f37 planemo upload for repository https://github.com/HegemanLab/w4mkmeans_galaxy_wrapper/tree/master commit 3e916537da6bb37e6f3927d7a11e98e0ab6ef5ec
eschen42
parents: 0
diff changeset
29 # On success of func(), tryCatchFunc produces
c415b7dc6f37 planemo upload for repository https://github.com/HegemanLab/w4mkmeans_galaxy_wrapper/tree/master commit 3e916537da6bb37e6f3927d7a11e98e0ab6ef5ec
eschen42
parents: 0
diff changeset
30 # list(success = TRUE, value = func(), msg = "")
c415b7dc6f37 planemo upload for repository https://github.com/HegemanLab/w4mkmeans_galaxy_wrapper/tree/master commit 3e916537da6bb37e6f3927d7a11e98e0ab6ef5ec
eschen42
parents: 0
diff changeset
31 # On failure of func(), tryCatchFunc produces
c415b7dc6f37 planemo upload for repository https://github.com/HegemanLab/w4mkmeans_galaxy_wrapper/tree/master commit 3e916537da6bb37e6f3927d7a11e98e0ab6ef5ec
eschen42
parents: 0
diff changeset
32 # list(success = FALSE, value = NA, msg = "the error message")
c415b7dc6f37 planemo upload for repository https://github.com/HegemanLab/w4mkmeans_galaxy_wrapper/tree/master commit 3e916537da6bb37e6f3927d7a11e98e0ab6ef5ec
eschen42
parents: 0
diff changeset
33 tryCatchFunc <- function(func) {
c415b7dc6f37 planemo upload for repository https://github.com/HegemanLab/w4mkmeans_galaxy_wrapper/tree/master commit 3e916537da6bb37e6f3927d7a11e98e0ab6ef5ec
eschen42
parents: 0
diff changeset
34 retval <- NULL
c415b7dc6f37 planemo upload for repository https://github.com/HegemanLab/w4mkmeans_galaxy_wrapper/tree/master commit 3e916537da6bb37e6f3927d7a11e98e0ab6ef5ec
eschen42
parents: 0
diff changeset
35 tryCatch(
c415b7dc6f37 planemo upload for repository https://github.com/HegemanLab/w4mkmeans_galaxy_wrapper/tree/master commit 3e916537da6bb37e6f3927d7a11e98e0ab6ef5ec
eschen42
parents: 0
diff changeset
36 expr = {
c415b7dc6f37 planemo upload for repository https://github.com/HegemanLab/w4mkmeans_galaxy_wrapper/tree/master commit 3e916537da6bb37e6f3927d7a11e98e0ab6ef5ec
eschen42
parents: 0
diff changeset
37 retval <- ( list( success = TRUE, value = func(), msg = "" ) )
c415b7dc6f37 planemo upload for repository https://github.com/HegemanLab/w4mkmeans_galaxy_wrapper/tree/master commit 3e916537da6bb37e6f3927d7a11e98e0ab6ef5ec
eschen42
parents: 0
diff changeset
38 }
c415b7dc6f37 planemo upload for repository https://github.com/HegemanLab/w4mkmeans_galaxy_wrapper/tree/master commit 3e916537da6bb37e6f3927d7a11e98e0ab6ef5ec
eschen42
parents: 0
diff changeset
39 , error = function(e) {
c415b7dc6f37 planemo upload for repository https://github.com/HegemanLab/w4mkmeans_galaxy_wrapper/tree/master commit 3e916537da6bb37e6f3927d7a11e98e0ab6ef5ec
eschen42
parents: 0
diff changeset
40 retval <<- list( success = FALSE, value = NA, msg = format_error(e) )
c415b7dc6f37 planemo upload for repository https://github.com/HegemanLab/w4mkmeans_galaxy_wrapper/tree/master commit 3e916537da6bb37e6f3927d7a11e98e0ab6ef5ec
eschen42
parents: 0
diff changeset
41 }
c415b7dc6f37 planemo upload for repository https://github.com/HegemanLab/w4mkmeans_galaxy_wrapper/tree/master commit 3e916537da6bb37e6f3927d7a11e98e0ab6ef5ec
eschen42
parents: 0
diff changeset
42 )
c415b7dc6f37 planemo upload for repository https://github.com/HegemanLab/w4mkmeans_galaxy_wrapper/tree/master commit 3e916537da6bb37e6f3927d7a11e98e0ab6ef5ec
eschen42
parents: 0
diff changeset
43 return (retval)
c415b7dc6f37 planemo upload for repository https://github.com/HegemanLab/w4mkmeans_galaxy_wrapper/tree/master commit 3e916537da6bb37e6f3927d7a11e98e0ab6ef5ec
eschen42
parents: 0
diff changeset
44 }
c415b7dc6f37 planemo upload for repository https://github.com/HegemanLab/w4mkmeans_galaxy_wrapper/tree/master commit 3e916537da6bb37e6f3927d7a11e98e0ab6ef5ec
eschen42
parents: 0
diff changeset
45
0
6ccbe18131a6 planemo upload for repository https://github.com/HegemanLab/w4mkmeans_galaxy_wrapper/tree/master commit 299e5c7fdb0d6eb0773f3660009f6d63c2082a8d
eschen42
parents:
diff changeset
46 # prepare.data.matrix - Prepare x.datamatrix for multivariate statistical analaysis (MVA)
6ccbe18131a6 planemo upload for repository https://github.com/HegemanLab/w4mkmeans_galaxy_wrapper/tree/master commit 299e5c7fdb0d6eb0773f3660009f6d63c2082a8d
eschen42
parents:
diff changeset
47 # - Motivation:
6ccbe18131a6 planemo upload for repository https://github.com/HegemanLab/w4mkmeans_galaxy_wrapper/tree/master commit 299e5c7fdb0d6eb0773f3660009f6d63c2082a8d
eschen42
parents:
diff changeset
48 # - Selection:
6ccbe18131a6 planemo upload for repository https://github.com/HegemanLab/w4mkmeans_galaxy_wrapper/tree/master commit 299e5c7fdb0d6eb0773f3660009f6d63c2082a8d
eschen42
parents:
diff changeset
49 # - You may want to exclude several samples from your analysis:
6ccbe18131a6 planemo upload for repository https://github.com/HegemanLab/w4mkmeans_galaxy_wrapper/tree/master commit 299e5c7fdb0d6eb0773f3660009f6d63c2082a8d
eschen42
parents:
diff changeset
50 # - If so, set the argument 'exclude.samples' to a vector of sample names
6ccbe18131a6 planemo upload for repository https://github.com/HegemanLab/w4mkmeans_galaxy_wrapper/tree/master commit 299e5c7fdb0d6eb0773f3660009f6d63c2082a8d
eschen42
parents:
diff changeset
51 # - You may want to exclude several features or features from your analysis:
6ccbe18131a6 planemo upload for repository https://github.com/HegemanLab/w4mkmeans_galaxy_wrapper/tree/master commit 299e5c7fdb0d6eb0773f3660009f6d63c2082a8d
eschen42
parents:
diff changeset
52 # - If so, set the argument 'exclude.features' to a vector of feature names
6ccbe18131a6 planemo upload for repository https://github.com/HegemanLab/w4mkmeans_galaxy_wrapper/tree/master commit 299e5c7fdb0d6eb0773f3660009f6d63c2082a8d
eschen42
parents:
diff changeset
53 # - Renaming samples:
6ccbe18131a6 planemo upload for repository https://github.com/HegemanLab/w4mkmeans_galaxy_wrapper/tree/master commit 299e5c7fdb0d6eb0773f3660009f6d63c2082a8d
eschen42
parents:
diff changeset
54 # - You may want to rename several samples from your analysis:
2
c415b7dc6f37 planemo upload for repository https://github.com/HegemanLab/w4mkmeans_galaxy_wrapper/tree/master commit 3e916537da6bb37e6f3927d7a11e98e0ab6ef5ec
eschen42
parents: 0
diff changeset
55 # - If so, set the argument 'sample.rename.function' to a function accepting a vector
0
6ccbe18131a6 planemo upload for repository https://github.com/HegemanLab/w4mkmeans_galaxy_wrapper/tree/master commit 299e5c7fdb0d6eb0773f3660009f6d63c2082a8d
eschen42
parents:
diff changeset
56 # of sample names and producing a vector of strings of equivalent length
6ccbe18131a6 planemo upload for repository https://github.com/HegemanLab/w4mkmeans_galaxy_wrapper/tree/master commit 299e5c7fdb0d6eb0773f3660009f6d63c2082a8d
eschen42
parents:
diff changeset
57 # - MVA is confounded by missing values.
6ccbe18131a6 planemo upload for repository https://github.com/HegemanLab/w4mkmeans_galaxy_wrapper/tree/master commit 299e5c7fdb0d6eb0773f3660009f6d63c2082a8d
eschen42
parents:
diff changeset
58 # - By default, this function imputes missing values as zero.
6ccbe18131a6 planemo upload for repository https://github.com/HegemanLab/w4mkmeans_galaxy_wrapper/tree/master commit 299e5c7fdb0d6eb0773f3660009f6d63c2082a8d
eschen42
parents:
diff changeset
59 # - For a different imputation, set the 'data.imputation' argument to a function
6ccbe18131a6 planemo upload for repository https://github.com/HegemanLab/w4mkmeans_galaxy_wrapper/tree/master commit 299e5c7fdb0d6eb0773f3660009f6d63c2082a8d
eschen42
parents:
diff changeset
60 # accepting a single matrix argument and returning a matrix of the same
6ccbe18131a6 planemo upload for repository https://github.com/HegemanLab/w4mkmeans_galaxy_wrapper/tree/master commit 299e5c7fdb0d6eb0773f3660009f6d63c2082a8d
eschen42
parents:
diff changeset
61 # dimensions as the argument.
6ccbe18131a6 planemo upload for repository https://github.com/HegemanLab/w4mkmeans_galaxy_wrapper/tree/master commit 299e5c7fdb0d6eb0773f3660009f6d63c2082a8d
eschen42
parents:
diff changeset
62 # - Transformation
6ccbe18131a6 planemo upload for repository https://github.com/HegemanLab/w4mkmeans_galaxy_wrapper/tree/master commit 299e5c7fdb0d6eb0773f3660009f6d63c2082a8d
eschen42
parents:
diff changeset
63 # - It may be desirable to transform the intensity data to reduce the range.
6ccbe18131a6 planemo upload for repository https://github.com/HegemanLab/w4mkmeans_galaxy_wrapper/tree/master commit 299e5c7fdb0d6eb0773f3660009f6d63c2082a8d
eschen42
parents:
diff changeset
64 # - By default, this function performs an eigth-root transformation:
6ccbe18131a6 planemo upload for repository https://github.com/HegemanLab/w4mkmeans_galaxy_wrapper/tree/master commit 299e5c7fdb0d6eb0773f3660009f6d63c2082a8d
eschen42
parents:
diff changeset
65 # - Any root-tranformation has the advantage of never being negative.
6ccbe18131a6 planemo upload for repository https://github.com/HegemanLab/w4mkmeans_galaxy_wrapper/tree/master commit 299e5c7fdb0d6eb0773f3660009f6d63c2082a8d
eschen42
parents:
diff changeset
66 # - Calculation of the eight-root is four times faster in my hands than log10.
2
c415b7dc6f37 planemo upload for repository https://github.com/HegemanLab/w4mkmeans_galaxy_wrapper/tree/master commit 3e916537da6bb37e6f3927d7a11e98e0ab6ef5ec
eschen42
parents: 0
diff changeset
67 # - However, it has the disadvantage that calculation of fold-differences
0
6ccbe18131a6 planemo upload for repository https://github.com/HegemanLab/w4mkmeans_galaxy_wrapper/tree/master commit 299e5c7fdb0d6eb0773f3660009f6d63c2082a8d
eschen42
parents:
diff changeset
68 # is not additive as with log-transformation.
6ccbe18131a6 planemo upload for repository https://github.com/HegemanLab/w4mkmeans_galaxy_wrapper/tree/master commit 299e5c7fdb0d6eb0773f3660009f6d63c2082a8d
eschen42
parents:
diff changeset
69 # - Rather, you must divide the values and raise to the eighth power.
6ccbe18131a6 planemo upload for repository https://github.com/HegemanLab/w4mkmeans_galaxy_wrapper/tree/master commit 299e5c7fdb0d6eb0773f3660009f6d63c2082a8d
eschen42
parents:
diff changeset
70 # - For a different transformation, set the 'data.transformation' argument
6ccbe18131a6 planemo upload for repository https://github.com/HegemanLab/w4mkmeans_galaxy_wrapper/tree/master commit 299e5c7fdb0d6eb0773f3660009f6d63c2082a8d
eschen42
parents:
diff changeset
71 # to a function accepting a single matrix argument.
6ccbe18131a6 planemo upload for repository https://github.com/HegemanLab/w4mkmeans_galaxy_wrapper/tree/master commit 299e5c7fdb0d6eb0773f3660009f6d63c2082a8d
eschen42
parents:
diff changeset
72 # - The function should be written to return a matrix of the same dimensions
6ccbe18131a6 planemo upload for repository https://github.com/HegemanLab/w4mkmeans_galaxy_wrapper/tree/master commit 299e5c7fdb0d6eb0773f3660009f6d63c2082a8d
eschen42
parents:
diff changeset
73 # as the argument.
6ccbe18131a6 planemo upload for repository https://github.com/HegemanLab/w4mkmeans_galaxy_wrapper/tree/master commit 299e5c7fdb0d6eb0773f3660009f6d63c2082a8d
eschen42
parents:
diff changeset
74 # arguments:
6ccbe18131a6 planemo upload for repository https://github.com/HegemanLab/w4mkmeans_galaxy_wrapper/tree/master commit 299e5c7fdb0d6eb0773f3660009f6d63c2082a8d
eschen42
parents:
diff changeset
75 # - x.matrix - matrix of intensities (or data.frame of sample metadata)
6ccbe18131a6 planemo upload for repository https://github.com/HegemanLab/w4mkmeans_galaxy_wrapper/tree/master commit 299e5c7fdb0d6eb0773f3660009f6d63c2082a8d
eschen42
parents:
diff changeset
76 # - one row per sample
6ccbe18131a6 planemo upload for repository https://github.com/HegemanLab/w4mkmeans_galaxy_wrapper/tree/master commit 299e5c7fdb0d6eb0773f3660009f6d63c2082a8d
eschen42
parents:
diff changeset
77 # - one column per feature or metadata attribute
6ccbe18131a6 planemo upload for repository https://github.com/HegemanLab/w4mkmeans_galaxy_wrapper/tree/master commit 299e5c7fdb0d6eb0773f3660009f6d63c2082a8d
eschen42
parents:
diff changeset
78 # - exclude.samples - vector of labels of matrix rows (samples) to omit from analysis
6ccbe18131a6 planemo upload for repository https://github.com/HegemanLab/w4mkmeans_galaxy_wrapper/tree/master commit 299e5c7fdb0d6eb0773f3660009f6d63c2082a8d
eschen42
parents:
diff changeset
79 # - exclude.features - vector of labels of matrix columnss (features) to omit from analysis
6ccbe18131a6 planemo upload for repository https://github.com/HegemanLab/w4mkmeans_galaxy_wrapper/tree/master commit 299e5c7fdb0d6eb0773f3660009f6d63c2082a8d
eschen42
parents:
diff changeset
80 # - sample.rename.function - function to be used to rename rows if necessary, or NULL
6ccbe18131a6 planemo upload for repository https://github.com/HegemanLab/w4mkmeans_galaxy_wrapper/tree/master commit 299e5c7fdb0d6eb0773f3660009f6d63c2082a8d
eschen42
parents:
diff changeset
81 # - e.g., sample.rename.function = function(x) {
6ccbe18131a6 planemo upload for repository https://github.com/HegemanLab/w4mkmeans_galaxy_wrapper/tree/master commit 299e5c7fdb0d6eb0773f3660009f6d63c2082a8d
eschen42
parents:
diff changeset
82 # sub("(.*)_.*","\\1", row.names(x))
6ccbe18131a6 planemo upload for repository https://github.com/HegemanLab/w4mkmeans_galaxy_wrapper/tree/master commit 299e5c7fdb0d6eb0773f3660009f6d63c2082a8d
eschen42
parents:
diff changeset
83 # }
6ccbe18131a6 planemo upload for repository https://github.com/HegemanLab/w4mkmeans_galaxy_wrapper/tree/master commit 299e5c7fdb0d6eb0773f3660009f6d63c2082a8d
eschen42
parents:
diff changeset
84 # - data.imputation - function applied to matrix to impute missing values
6ccbe18131a6 planemo upload for repository https://github.com/HegemanLab/w4mkmeans_galaxy_wrapper/tree/master commit 299e5c7fdb0d6eb0773f3660009f6d63c2082a8d
eschen42
parents:
diff changeset
85 # - e.g., data.imputation = function(m) {
6ccbe18131a6 planemo upload for repository https://github.com/HegemanLab/w4mkmeans_galaxy_wrapper/tree/master commit 299e5c7fdb0d6eb0773f3660009f6d63c2082a8d
eschen42
parents:
diff changeset
86 # m[is.na(m)] <- min(m, na.rm = TRUE) / 100
6ccbe18131a6 planemo upload for repository https://github.com/HegemanLab/w4mkmeans_galaxy_wrapper/tree/master commit 299e5c7fdb0d6eb0773f3660009f6d63c2082a8d
eschen42
parents:
diff changeset
87 # return (m)
6ccbe18131a6 planemo upload for repository https://github.com/HegemanLab/w4mkmeans_galaxy_wrapper/tree/master commit 299e5c7fdb0d6eb0773f3660009f6d63c2082a8d
eschen42
parents:
diff changeset
88 # }
6ccbe18131a6 planemo upload for repository https://github.com/HegemanLab/w4mkmeans_galaxy_wrapper/tree/master commit 299e5c7fdb0d6eb0773f3660009f6d63c2082a8d
eschen42
parents:
diff changeset
89 # - data.transformation - function applied to matrix cells
6ccbe18131a6 planemo upload for repository https://github.com/HegemanLab/w4mkmeans_galaxy_wrapper/tree/master commit 299e5c7fdb0d6eb0773f3660009f6d63c2082a8d
eschen42
parents:
diff changeset
90 # - e.g., data.transformation = function(x) { return( log10(x) ) }
6ccbe18131a6 planemo upload for repository https://github.com/HegemanLab/w4mkmeans_galaxy_wrapper/tree/master commit 299e5c7fdb0d6eb0773f3660009f6d63c2082a8d
eschen42
parents:
diff changeset
91 # or, data.transformation = log10
6ccbe18131a6 planemo upload for repository https://github.com/HegemanLab/w4mkmeans_galaxy_wrapper/tree/master commit 299e5c7fdb0d6eb0773f3660009f6d63c2082a8d
eschen42
parents:
diff changeset
92 # result value:
6ccbe18131a6 planemo upload for repository https://github.com/HegemanLab/w4mkmeans_galaxy_wrapper/tree/master commit 299e5c7fdb0d6eb0773f3660009f6d63c2082a8d
eschen42
parents:
diff changeset
93 # transformed, imputed x.datamatrix with renamed rows and with neither excluded values nor features
6ccbe18131a6 planemo upload for repository https://github.com/HegemanLab/w4mkmeans_galaxy_wrapper/tree/master commit 299e5c7fdb0d6eb0773f3660009f6d63c2082a8d
eschen42
parents:
diff changeset
94 #
6ccbe18131a6 planemo upload for repository https://github.com/HegemanLab/w4mkmeans_galaxy_wrapper/tree/master commit 299e5c7fdb0d6eb0773f3660009f6d63c2082a8d
eschen42
parents:
diff changeset
95 ################################
6ccbe18131a6 planemo upload for repository https://github.com/HegemanLab/w4mkmeans_galaxy_wrapper/tree/master commit 299e5c7fdb0d6eb0773f3660009f6d63c2082a8d
eschen42
parents:
diff changeset
96 ##
6ccbe18131a6 planemo upload for repository https://github.com/HegemanLab/w4mkmeans_galaxy_wrapper/tree/master commit 299e5c7fdb0d6eb0773f3660009f6d63c2082a8d
eschen42
parents:
diff changeset
97 ## Notes regarding the effectiveness and performance of the data transformation method.
6ccbe18131a6 planemo upload for repository https://github.com/HegemanLab/w4mkmeans_galaxy_wrapper/tree/master commit 299e5c7fdb0d6eb0773f3660009f6d63c2082a8d
eschen42
parents:
diff changeset
98 ##
6ccbe18131a6 planemo upload for repository https://github.com/HegemanLab/w4mkmeans_galaxy_wrapper/tree/master commit 299e5c7fdb0d6eb0773f3660009f6d63c2082a8d
eschen42
parents:
diff changeset
99 ## The two transformations that I tried (log10 and 8th root) required different imputation methods.
6ccbe18131a6 planemo upload for repository https://github.com/HegemanLab/w4mkmeans_galaxy_wrapper/tree/master commit 299e5c7fdb0d6eb0773f3660009f6d63c2082a8d
eschen42
parents:
diff changeset
100 ##
6ccbe18131a6 planemo upload for repository https://github.com/HegemanLab/w4mkmeans_galaxy_wrapper/tree/master commit 299e5c7fdb0d6eb0773f3660009f6d63c2082a8d
eschen42
parents:
diff changeset
101 ## For the LCMS resin data set that I was working with, separation in MVA was nearly equivalent for:
6ccbe18131a6 planemo upload for repository https://github.com/HegemanLab/w4mkmeans_galaxy_wrapper/tree/master commit 299e5c7fdb0d6eb0773f3660009f6d63c2082a8d
eschen42
parents:
diff changeset
102 ## data.imputation <- function(x.matrix) {
6ccbe18131a6 planemo upload for repository https://github.com/HegemanLab/w4mkmeans_galaxy_wrapper/tree/master commit 299e5c7fdb0d6eb0773f3660009f6d63c2082a8d
eschen42
parents:
diff changeset
103 ## x.matrix[is.na(x.matrix)] <- 0
6ccbe18131a6 planemo upload for repository https://github.com/HegemanLab/w4mkmeans_galaxy_wrapper/tree/master commit 299e5c7fdb0d6eb0773f3660009f6d63c2082a8d
eschen42
parents:
diff changeset
104 ## return (x.matrix)
6ccbe18131a6 planemo upload for repository https://github.com/HegemanLab/w4mkmeans_galaxy_wrapper/tree/master commit 299e5c7fdb0d6eb0773f3660009f6d63c2082a8d
eschen42
parents:
diff changeset
105 ## }
6ccbe18131a6 planemo upload for repository https://github.com/HegemanLab/w4mkmeans_galaxy_wrapper/tree/master commit 299e5c7fdb0d6eb0773f3660009f6d63c2082a8d
eschen42
parents:
diff changeset
106 ## data.transformation <- function(x) {
6ccbe18131a6 planemo upload for repository https://github.com/HegemanLab/w4mkmeans_galaxy_wrapper/tree/master commit 299e5c7fdb0d6eb0773f3660009f6d63c2082a8d
eschen42
parents:
diff changeset
107 ## sqrt( sqrt( sqrt(x) ) )
6ccbe18131a6 planemo upload for repository https://github.com/HegemanLab/w4mkmeans_galaxy_wrapper/tree/master commit 299e5c7fdb0d6eb0773f3660009f6d63c2082a8d
eschen42
parents:
diff changeset
108 ## }
6ccbe18131a6 planemo upload for repository https://github.com/HegemanLab/w4mkmeans_galaxy_wrapper/tree/master commit 299e5c7fdb0d6eb0773f3660009f6d63c2082a8d
eschen42
parents:
diff changeset
109 ## and
6ccbe18131a6 planemo upload for repository https://github.com/HegemanLab/w4mkmeans_galaxy_wrapper/tree/master commit 299e5c7fdb0d6eb0773f3660009f6d63c2082a8d
eschen42
parents:
diff changeset
110 ## data.imputation <- function(x.matrix) {
6ccbe18131a6 planemo upload for repository https://github.com/HegemanLab/w4mkmeans_galaxy_wrapper/tree/master commit 299e5c7fdb0d6eb0773f3660009f6d63c2082a8d
eschen42
parents:
diff changeset
111 ## x.matrix[is.na(x.matrix)] <- min(x.matrix, na.rm = TRUE) / 100
6ccbe18131a6 planemo upload for repository https://github.com/HegemanLab/w4mkmeans_galaxy_wrapper/tree/master commit 299e5c7fdb0d6eb0773f3660009f6d63c2082a8d
eschen42
parents:
diff changeset
112 ## return (x.matrix)
6ccbe18131a6 planemo upload for repository https://github.com/HegemanLab/w4mkmeans_galaxy_wrapper/tree/master commit 299e5c7fdb0d6eb0773f3660009f6d63c2082a8d
eschen42
parents:
diff changeset
113 ## }
6ccbe18131a6 planemo upload for repository https://github.com/HegemanLab/w4mkmeans_galaxy_wrapper/tree/master commit 299e5c7fdb0d6eb0773f3660009f6d63c2082a8d
eschen42
parents:
diff changeset
114 ## data.transformation <- function(x) {
6ccbe18131a6 planemo upload for repository https://github.com/HegemanLab/w4mkmeans_galaxy_wrapper/tree/master commit 299e5c7fdb0d6eb0773f3660009f6d63c2082a8d
eschen42
parents:
diff changeset
115 ## log10(x)
6ccbe18131a6 planemo upload for repository https://github.com/HegemanLab/w4mkmeans_galaxy_wrapper/tree/master commit 299e5c7fdb0d6eb0773f3660009f6d63c2082a8d
eschen42
parents:
diff changeset
116 ## }
6ccbe18131a6 planemo upload for repository https://github.com/HegemanLab/w4mkmeans_galaxy_wrapper/tree/master commit 299e5c7fdb0d6eb0773f3660009f6d63c2082a8d
eschen42
parents:
diff changeset
117 ##
6ccbe18131a6 planemo upload for repository https://github.com/HegemanLab/w4mkmeans_galaxy_wrapper/tree/master commit 299e5c7fdb0d6eb0773f3660009f6d63c2082a8d
eschen42
parents:
diff changeset
118 ## Note further that triple application of the square root:
6ccbe18131a6 planemo upload for repository https://github.com/HegemanLab/w4mkmeans_galaxy_wrapper/tree/master commit 299e5c7fdb0d6eb0773f3660009f6d63c2082a8d
eschen42
parents:
diff changeset
119 ## - may be four times faster than log10:
6ccbe18131a6 planemo upload for repository https://github.com/HegemanLab/w4mkmeans_galaxy_wrapper/tree/master commit 299e5c7fdb0d6eb0773f3660009f6d63c2082a8d
eschen42
parents:
diff changeset
120 ## - may be three times faster than log2:
6ccbe18131a6 planemo upload for repository https://github.com/HegemanLab/w4mkmeans_galaxy_wrapper/tree/master commit 299e5c7fdb0d6eb0773f3660009f6d63c2082a8d
eschen42
parents:
diff changeset
121 ##
6ccbe18131a6 planemo upload for repository https://github.com/HegemanLab/w4mkmeans_galaxy_wrapper/tree/master commit 299e5c7fdb0d6eb0773f3660009f6d63c2082a8d
eschen42
parents:
diff changeset
122 ## system.time( junk <- sqrt( sqrt( sqrt(1:100000000) ) ) )
6ccbe18131a6 planemo upload for repository https://github.com/HegemanLab/w4mkmeans_galaxy_wrapper/tree/master commit 299e5c7fdb0d6eb0773f3660009f6d63c2082a8d
eschen42
parents:
diff changeset
123 ## user system elapsed
6ccbe18131a6 planemo upload for repository https://github.com/HegemanLab/w4mkmeans_galaxy_wrapper/tree/master commit 299e5c7fdb0d6eb0773f3660009f6d63c2082a8d
eschen42
parents:
diff changeset
124 ## 0.832 0.236 1.069
6ccbe18131a6 planemo upload for repository https://github.com/HegemanLab/w4mkmeans_galaxy_wrapper/tree/master commit 299e5c7fdb0d6eb0773f3660009f6d63c2082a8d
eschen42
parents:
diff changeset
125 ## system.time( junk <- log10(1:100000000) )
6ccbe18131a6 planemo upload for repository https://github.com/HegemanLab/w4mkmeans_galaxy_wrapper/tree/master commit 299e5c7fdb0d6eb0773f3660009f6d63c2082a8d
eschen42
parents:
diff changeset
126 ## user system elapsed
6ccbe18131a6 planemo upload for repository https://github.com/HegemanLab/w4mkmeans_galaxy_wrapper/tree/master commit 299e5c7fdb0d6eb0773f3660009f6d63c2082a8d
eschen42
parents:
diff changeset
127 ## 3.936 0.400 4.337
6ccbe18131a6 planemo upload for repository https://github.com/HegemanLab/w4mkmeans_galaxy_wrapper/tree/master commit 299e5c7fdb0d6eb0773f3660009f6d63c2082a8d
eschen42
parents:
diff changeset
128 ## system.time( junk <- log2(1:100000000) )
6ccbe18131a6 planemo upload for repository https://github.com/HegemanLab/w4mkmeans_galaxy_wrapper/tree/master commit 299e5c7fdb0d6eb0773f3660009f6d63c2082a8d
eschen42
parents:
diff changeset
129 ## user system elapsed
6ccbe18131a6 planemo upload for repository https://github.com/HegemanLab/w4mkmeans_galaxy_wrapper/tree/master commit 299e5c7fdb0d6eb0773f3660009f6d63c2082a8d
eschen42
parents:
diff changeset
130 ## 2.784 0.320 3.101
6ccbe18131a6 planemo upload for repository https://github.com/HegemanLab/w4mkmeans_galaxy_wrapper/tree/master commit 299e5c7fdb0d6eb0773f3660009f6d63c2082a8d
eschen42
parents:
diff changeset
131 ##
6ccbe18131a6 planemo upload for repository https://github.com/HegemanLab/w4mkmeans_galaxy_wrapper/tree/master commit 299e5c7fdb0d6eb0773f3660009f6d63c2082a8d
eschen42
parents:
diff changeset
132 ################################
6ccbe18131a6 planemo upload for repository https://github.com/HegemanLab/w4mkmeans_galaxy_wrapper/tree/master commit 299e5c7fdb0d6eb0773f3660009f6d63c2082a8d
eschen42
parents:
diff changeset
133 #
6ccbe18131a6 planemo upload for repository https://github.com/HegemanLab/w4mkmeans_galaxy_wrapper/tree/master commit 299e5c7fdb0d6eb0773f3660009f6d63c2082a8d
eschen42
parents:
diff changeset
134 prepare.data.matrix <- function(
6ccbe18131a6 planemo upload for repository https://github.com/HegemanLab/w4mkmeans_galaxy_wrapper/tree/master commit 299e5c7fdb0d6eb0773f3660009f6d63c2082a8d
eschen42
parents:
diff changeset
135 x.matrix
6ccbe18131a6 planemo upload for repository https://github.com/HegemanLab/w4mkmeans_galaxy_wrapper/tree/master commit 299e5c7fdb0d6eb0773f3660009f6d63c2082a8d
eschen42
parents:
diff changeset
136 , exclude.samples = NULL
6ccbe18131a6 planemo upload for repository https://github.com/HegemanLab/w4mkmeans_galaxy_wrapper/tree/master commit 299e5c7fdb0d6eb0773f3660009f6d63c2082a8d
eschen42
parents:
diff changeset
137 , exclude.features = NULL
6ccbe18131a6 planemo upload for repository https://github.com/HegemanLab/w4mkmeans_galaxy_wrapper/tree/master commit 299e5c7fdb0d6eb0773f3660009f6d63c2082a8d
eschen42
parents:
diff changeset
138 , sample.rename.function = NULL
6ccbe18131a6 planemo upload for repository https://github.com/HegemanLab/w4mkmeans_galaxy_wrapper/tree/master commit 299e5c7fdb0d6eb0773f3660009f6d63c2082a8d
eschen42
parents:
diff changeset
139 , data.imputation =
6ccbe18131a6 planemo upload for repository https://github.com/HegemanLab/w4mkmeans_galaxy_wrapper/tree/master commit 299e5c7fdb0d6eb0773f3660009f6d63c2082a8d
eschen42
parents:
diff changeset
140 function(m) {
6ccbe18131a6 planemo upload for repository https://github.com/HegemanLab/w4mkmeans_galaxy_wrapper/tree/master commit 299e5c7fdb0d6eb0773f3660009f6d63c2082a8d
eschen42
parents:
diff changeset
141 # replace NA values with zero
6ccbe18131a6 planemo upload for repository https://github.com/HegemanLab/w4mkmeans_galaxy_wrapper/tree/master commit 299e5c7fdb0d6eb0773f3660009f6d63c2082a8d
eschen42
parents:
diff changeset
142 m[is.na(m)] <- 0
6ccbe18131a6 planemo upload for repository https://github.com/HegemanLab/w4mkmeans_galaxy_wrapper/tree/master commit 299e5c7fdb0d6eb0773f3660009f6d63c2082a8d
eschen42
parents:
diff changeset
143 # replace negative values with zero, if applicable (It should never be applicable!)
6ccbe18131a6 planemo upload for repository https://github.com/HegemanLab/w4mkmeans_galaxy_wrapper/tree/master commit 299e5c7fdb0d6eb0773f3660009f6d63c2082a8d
eschen42
parents:
diff changeset
144 if (min(m < 0)) {
6ccbe18131a6 planemo upload for repository https://github.com/HegemanLab/w4mkmeans_galaxy_wrapper/tree/master commit 299e5c7fdb0d6eb0773f3660009f6d63c2082a8d
eschen42
parents:
diff changeset
145 m <- matrix(lapply(X = m, FUN = function(z) {max(z,0)}), nrow = nrow(m) )
6ccbe18131a6 planemo upload for repository https://github.com/HegemanLab/w4mkmeans_galaxy_wrapper/tree/master commit 299e5c7fdb0d6eb0773f3660009f6d63c2082a8d
eschen42
parents:
diff changeset
146 }
6ccbe18131a6 planemo upload for repository https://github.com/HegemanLab/w4mkmeans_galaxy_wrapper/tree/master commit 299e5c7fdb0d6eb0773f3660009f6d63c2082a8d
eschen42
parents:
diff changeset
147 # return matrix as the result
6ccbe18131a6 planemo upload for repository https://github.com/HegemanLab/w4mkmeans_galaxy_wrapper/tree/master commit 299e5c7fdb0d6eb0773f3660009f6d63c2082a8d
eschen42
parents:
diff changeset
148 return (m)
6ccbe18131a6 planemo upload for repository https://github.com/HegemanLab/w4mkmeans_galaxy_wrapper/tree/master commit 299e5c7fdb0d6eb0773f3660009f6d63c2082a8d
eschen42
parents:
diff changeset
149 }
6ccbe18131a6 planemo upload for repository https://github.com/HegemanLab/w4mkmeans_galaxy_wrapper/tree/master commit 299e5c7fdb0d6eb0773f3660009f6d63c2082a8d
eschen42
parents:
diff changeset
150 , data.transformation = function(x) {
6ccbe18131a6 planemo upload for repository https://github.com/HegemanLab/w4mkmeans_galaxy_wrapper/tree/master commit 299e5c7fdb0d6eb0773f3660009f6d63c2082a8d
eschen42
parents:
diff changeset
151 sqrt( sqrt( sqrt(x) ) )
6ccbe18131a6 planemo upload for repository https://github.com/HegemanLab/w4mkmeans_galaxy_wrapper/tree/master commit 299e5c7fdb0d6eb0773f3660009f6d63c2082a8d
eschen42
parents:
diff changeset
152 }
6ccbe18131a6 planemo upload for repository https://github.com/HegemanLab/w4mkmeans_galaxy_wrapper/tree/master commit 299e5c7fdb0d6eb0773f3660009f6d63c2082a8d
eschen42
parents:
diff changeset
153 , en = new.env()
6ccbe18131a6 planemo upload for repository https://github.com/HegemanLab/w4mkmeans_galaxy_wrapper/tree/master commit 299e5c7fdb0d6eb0773f3660009f6d63c2082a8d
eschen42
parents:
diff changeset
154 ) {
2
c415b7dc6f37 planemo upload for repository https://github.com/HegemanLab/w4mkmeans_galaxy_wrapper/tree/master commit 3e916537da6bb37e6f3927d7a11e98e0ab6ef5ec
eschen42
parents: 0
diff changeset
155 # log to environment
c415b7dc6f37 planemo upload for repository https://github.com/HegemanLab/w4mkmeans_galaxy_wrapper/tree/master commit 3e916537da6bb37e6f3927d7a11e98e0ab6ef5ec
eschen42
parents: 0
diff changeset
156 if ( !exists("log", envir = en) ) {
c415b7dc6f37 planemo upload for repository https://github.com/HegemanLab/w4mkmeans_galaxy_wrapper/tree/master commit 3e916537da6bb37e6f3927d7a11e98e0ab6ef5ec
eschen42
parents: 0
diff changeset
157 en$log <- c()
c415b7dc6f37 planemo upload for repository https://github.com/HegemanLab/w4mkmeans_galaxy_wrapper/tree/master commit 3e916537da6bb37e6f3927d7a11e98e0ab6ef5ec
eschen42
parents: 0
diff changeset
158 }
c415b7dc6f37 planemo upload for repository https://github.com/HegemanLab/w4mkmeans_galaxy_wrapper/tree/master commit 3e916537da6bb37e6f3927d7a11e98e0ab6ef5ec
eschen42
parents: 0
diff changeset
159 enlog <- function(s) { en$log <- c(en$log, s); s }
c415b7dc6f37 planemo upload for repository https://github.com/HegemanLab/w4mkmeans_galaxy_wrapper/tree/master commit 3e916537da6bb37e6f3927d7a11e98e0ab6ef5ec
eschen42
parents: 0
diff changeset
160 #enlog("foo")
c415b7dc6f37 planemo upload for repository https://github.com/HegemanLab/w4mkmeans_galaxy_wrapper/tree/master commit 3e916537da6bb37e6f3927d7a11e98e0ab6ef5ec
eschen42
parents: 0
diff changeset
161
0
6ccbe18131a6 planemo upload for repository https://github.com/HegemanLab/w4mkmeans_galaxy_wrapper/tree/master commit 299e5c7fdb0d6eb0773f3660009f6d63c2082a8d
eschen42
parents:
diff changeset
162 # MatVar - Compute variance of rows or columns of a matrix
6ccbe18131a6 planemo upload for repository https://github.com/HegemanLab/w4mkmeans_galaxy_wrapper/tree/master commit 299e5c7fdb0d6eb0773f3660009f6d63c2082a8d
eschen42
parents:
diff changeset
163 # ref: http://stackoverflow.com/a/25100036
6ccbe18131a6 planemo upload for repository https://github.com/HegemanLab/w4mkmeans_galaxy_wrapper/tree/master commit 299e5c7fdb0d6eb0773f3660009f6d63c2082a8d
eschen42
parents:
diff changeset
164 # For row variance, dim == 1, for col variance, dim == 2
6ccbe18131a6 planemo upload for repository https://github.com/HegemanLab/w4mkmeans_galaxy_wrapper/tree/master commit 299e5c7fdb0d6eb0773f3660009f6d63c2082a8d
eschen42
parents:
diff changeset
165 MatVar <- function(x, dim = 1) {
6ccbe18131a6 planemo upload for repository https://github.com/HegemanLab/w4mkmeans_galaxy_wrapper/tree/master commit 299e5c7fdb0d6eb0773f3660009f6d63c2082a8d
eschen42
parents:
diff changeset
166 if (dim == 1) {
6ccbe18131a6 planemo upload for repository https://github.com/HegemanLab/w4mkmeans_galaxy_wrapper/tree/master commit 299e5c7fdb0d6eb0773f3660009f6d63c2082a8d
eschen42
parents:
diff changeset
167 dim.x.2 <- dim(x)[2]
6ccbe18131a6 planemo upload for repository https://github.com/HegemanLab/w4mkmeans_galaxy_wrapper/tree/master commit 299e5c7fdb0d6eb0773f3660009f6d63c2082a8d
eschen42
parents:
diff changeset
168 if ( dim.x.2 == 0 )
6ccbe18131a6 planemo upload for repository https://github.com/HegemanLab/w4mkmeans_galaxy_wrapper/tree/master commit 299e5c7fdb0d6eb0773f3660009f6d63c2082a8d
eschen42
parents:
diff changeset
169 stop("MatVar: there are zero columns")
6ccbe18131a6 planemo upload for repository https://github.com/HegemanLab/w4mkmeans_galaxy_wrapper/tree/master commit 299e5c7fdb0d6eb0773f3660009f6d63c2082a8d
eschen42
parents:
diff changeset
170 if ( dim.x.2 == 1 ) {
6ccbe18131a6 planemo upload for repository https://github.com/HegemanLab/w4mkmeans_galaxy_wrapper/tree/master commit 299e5c7fdb0d6eb0773f3660009f6d63c2082a8d
eschen42
parents:
diff changeset
171 stop("MatVar: a single column is insufficient to calculate a variance")
6ccbe18131a6 planemo upload for repository https://github.com/HegemanLab/w4mkmeans_galaxy_wrapper/tree/master commit 299e5c7fdb0d6eb0773f3660009f6d63c2082a8d
eschen42
parents:
diff changeset
172 # return ( rep.int(x = 0, times = nrow(x)) )
6ccbe18131a6 planemo upload for repository https://github.com/HegemanLab/w4mkmeans_galaxy_wrapper/tree/master commit 299e5c7fdb0d6eb0773f3660009f6d63c2082a8d
eschen42
parents:
diff changeset
173 } else {
6ccbe18131a6 planemo upload for repository https://github.com/HegemanLab/w4mkmeans_galaxy_wrapper/tree/master commit 299e5c7fdb0d6eb0773f3660009f6d63c2082a8d
eschen42
parents:
diff changeset
174 return ( rowSums( (x - rowMeans(x))^2 ) / ( dim(x)[2] - 1 ) )
6ccbe18131a6 planemo upload for repository https://github.com/HegemanLab/w4mkmeans_galaxy_wrapper/tree/master commit 299e5c7fdb0d6eb0773f3660009f6d63c2082a8d
eschen42
parents:
diff changeset
175 }
6ccbe18131a6 planemo upload for repository https://github.com/HegemanLab/w4mkmeans_galaxy_wrapper/tree/master commit 299e5c7fdb0d6eb0773f3660009f6d63c2082a8d
eschen42
parents:
diff changeset
176 } else if (dim == 2) {
6ccbe18131a6 planemo upload for repository https://github.com/HegemanLab/w4mkmeans_galaxy_wrapper/tree/master commit 299e5c7fdb0d6eb0773f3660009f6d63c2082a8d
eschen42
parents:
diff changeset
177 dim.x.1 <- dim(x)[1]
6ccbe18131a6 planemo upload for repository https://github.com/HegemanLab/w4mkmeans_galaxy_wrapper/tree/master commit 299e5c7fdb0d6eb0773f3660009f6d63c2082a8d
eschen42
parents:
diff changeset
178 if ( dim.x.1 == 0 ) {
6ccbe18131a6 planemo upload for repository https://github.com/HegemanLab/w4mkmeans_galaxy_wrapper/tree/master commit 299e5c7fdb0d6eb0773f3660009f6d63c2082a8d
eschen42
parents:
diff changeset
179 stop("MatVar: there are zero rows")
6ccbe18131a6 planemo upload for repository https://github.com/HegemanLab/w4mkmeans_galaxy_wrapper/tree/master commit 299e5c7fdb0d6eb0773f3660009f6d63c2082a8d
eschen42
parents:
diff changeset
180 }
6ccbe18131a6 planemo upload for repository https://github.com/HegemanLab/w4mkmeans_galaxy_wrapper/tree/master commit 299e5c7fdb0d6eb0773f3660009f6d63c2082a8d
eschen42
parents:
diff changeset
181 if ( dim.x.1 == 1 ) {
6ccbe18131a6 planemo upload for repository https://github.com/HegemanLab/w4mkmeans_galaxy_wrapper/tree/master commit 299e5c7fdb0d6eb0773f3660009f6d63c2082a8d
eschen42
parents:
diff changeset
182 stop("MatVar: a single row is insufficient to calculate a variance")
6ccbe18131a6 planemo upload for repository https://github.com/HegemanLab/w4mkmeans_galaxy_wrapper/tree/master commit 299e5c7fdb0d6eb0773f3660009f6d63c2082a8d
eschen42
parents:
diff changeset
183 # return ( rep.int(x = 0, times = ncol(x)) )
6ccbe18131a6 planemo upload for repository https://github.com/HegemanLab/w4mkmeans_galaxy_wrapper/tree/master commit 299e5c7fdb0d6eb0773f3660009f6d63c2082a8d
eschen42
parents:
diff changeset
184 } else {
6ccbe18131a6 planemo upload for repository https://github.com/HegemanLab/w4mkmeans_galaxy_wrapper/tree/master commit 299e5c7fdb0d6eb0773f3660009f6d63c2082a8d
eschen42
parents:
diff changeset
185 return ( rowSums( (t(x) - colMeans(x))^2 ) / ( dim(x)[1] - 1 ) )
6ccbe18131a6 planemo upload for repository https://github.com/HegemanLab/w4mkmeans_galaxy_wrapper/tree/master commit 299e5c7fdb0d6eb0773f3660009f6d63c2082a8d
eschen42
parents:
diff changeset
186 }
6ccbe18131a6 planemo upload for repository https://github.com/HegemanLab/w4mkmeans_galaxy_wrapper/tree/master commit 299e5c7fdb0d6eb0773f3660009f6d63c2082a8d
eschen42
parents:
diff changeset
187 } else stop("Please enter valid dimension, for rows, dim = 1; for colums, dim = 2")
6ccbe18131a6 planemo upload for repository https://github.com/HegemanLab/w4mkmeans_galaxy_wrapper/tree/master commit 299e5c7fdb0d6eb0773f3660009f6d63c2082a8d
eschen42
parents:
diff changeset
188 }
6ccbe18131a6 planemo upload for repository https://github.com/HegemanLab/w4mkmeans_galaxy_wrapper/tree/master commit 299e5c7fdb0d6eb0773f3660009f6d63c2082a8d
eschen42
parents:
diff changeset
189
6ccbe18131a6 planemo upload for repository https://github.com/HegemanLab/w4mkmeans_galaxy_wrapper/tree/master commit 299e5c7fdb0d6eb0773f3660009f6d63c2082a8d
eschen42
parents:
diff changeset
190 nonzero.var <- function(x) {
6ccbe18131a6 planemo upload for repository https://github.com/HegemanLab/w4mkmeans_galaxy_wrapper/tree/master commit 299e5c7fdb0d6eb0773f3660009f6d63c2082a8d
eschen42
parents:
diff changeset
191 if (nrow(x) == 0) {
6ccbe18131a6 planemo upload for repository https://github.com/HegemanLab/w4mkmeans_galaxy_wrapper/tree/master commit 299e5c7fdb0d6eb0773f3660009f6d63c2082a8d
eschen42
parents:
diff changeset
192 stop("matrix has no rows")
6ccbe18131a6 planemo upload for repository https://github.com/HegemanLab/w4mkmeans_galaxy_wrapper/tree/master commit 299e5c7fdb0d6eb0773f3660009f6d63c2082a8d
eschen42
parents:
diff changeset
193 }
6ccbe18131a6 planemo upload for repository https://github.com/HegemanLab/w4mkmeans_galaxy_wrapper/tree/master commit 299e5c7fdb0d6eb0773f3660009f6d63c2082a8d
eschen42
parents:
diff changeset
194 if (ncol(x) == 0) {
6ccbe18131a6 planemo upload for repository https://github.com/HegemanLab/w4mkmeans_galaxy_wrapper/tree/master commit 299e5c7fdb0d6eb0773f3660009f6d63c2082a8d
eschen42
parents:
diff changeset
195 stop("matrix has no columns")
6ccbe18131a6 planemo upload for repository https://github.com/HegemanLab/w4mkmeans_galaxy_wrapper/tree/master commit 299e5c7fdb0d6eb0773f3660009f6d63c2082a8d
eschen42
parents:
diff changeset
196 }
6ccbe18131a6 planemo upload for repository https://github.com/HegemanLab/w4mkmeans_galaxy_wrapper/tree/master commit 299e5c7fdb0d6eb0773f3660009f6d63c2082a8d
eschen42
parents:
diff changeset
197 if ( is.numeric(x) ) {
6ccbe18131a6 planemo upload for repository https://github.com/HegemanLab/w4mkmeans_galaxy_wrapper/tree/master commit 299e5c7fdb0d6eb0773f3660009f6d63c2082a8d
eschen42
parents:
diff changeset
198 # exclude any rows with zero variance
6ccbe18131a6 planemo upload for repository https://github.com/HegemanLab/w4mkmeans_galaxy_wrapper/tree/master commit 299e5c7fdb0d6eb0773f3660009f6d63c2082a8d
eschen42
parents:
diff changeset
199 row.vars <- MatVar(x, dim = 1)
6ccbe18131a6 planemo upload for repository https://github.com/HegemanLab/w4mkmeans_galaxy_wrapper/tree/master commit 299e5c7fdb0d6eb0773f3660009f6d63c2082a8d
eschen42
parents:
diff changeset
200 nonzero.row.vars <- row.vars > 0
6ccbe18131a6 planemo upload for repository https://github.com/HegemanLab/w4mkmeans_galaxy_wrapper/tree/master commit 299e5c7fdb0d6eb0773f3660009f6d63c2082a8d
eschen42
parents:
diff changeset
201 nonzero.rows <- row.vars[nonzero.row.vars]
6ccbe18131a6 planemo upload for repository https://github.com/HegemanLab/w4mkmeans_galaxy_wrapper/tree/master commit 299e5c7fdb0d6eb0773f3660009f6d63c2082a8d
eschen42
parents:
diff changeset
202 if ( length(rownames(x)) != length(rownames(nonzero.rows)) ) {
6ccbe18131a6 planemo upload for repository https://github.com/HegemanLab/w4mkmeans_galaxy_wrapper/tree/master commit 299e5c7fdb0d6eb0773f3660009f6d63c2082a8d
eschen42
parents:
diff changeset
203 row.names <- attr(nonzero.rows,"names")
6ccbe18131a6 planemo upload for repository https://github.com/HegemanLab/w4mkmeans_galaxy_wrapper/tree/master commit 299e5c7fdb0d6eb0773f3660009f6d63c2082a8d
eschen42
parents:
diff changeset
204 x <- x[ row.names, , drop = FALSE ]
6ccbe18131a6 planemo upload for repository https://github.com/HegemanLab/w4mkmeans_galaxy_wrapper/tree/master commit 299e5c7fdb0d6eb0773f3660009f6d63c2082a8d
eschen42
parents:
diff changeset
205 }
2
c415b7dc6f37 planemo upload for repository https://github.com/HegemanLab/w4mkmeans_galaxy_wrapper/tree/master commit 3e916537da6bb37e6f3927d7a11e98e0ab6ef5ec
eschen42
parents: 0
diff changeset
206
0
6ccbe18131a6 planemo upload for repository https://github.com/HegemanLab/w4mkmeans_galaxy_wrapper/tree/master commit 299e5c7fdb0d6eb0773f3660009f6d63c2082a8d
eschen42
parents:
diff changeset
207 # exclude any columns with zero variance
6ccbe18131a6 planemo upload for repository https://github.com/HegemanLab/w4mkmeans_galaxy_wrapper/tree/master commit 299e5c7fdb0d6eb0773f3660009f6d63c2082a8d
eschen42
parents:
diff changeset
208 column.vars <- MatVar(x, dim = 2)
6ccbe18131a6 planemo upload for repository https://github.com/HegemanLab/w4mkmeans_galaxy_wrapper/tree/master commit 299e5c7fdb0d6eb0773f3660009f6d63c2082a8d
eschen42
parents:
diff changeset
209 nonzero.column.vars <- column.vars > 0
6ccbe18131a6 planemo upload for repository https://github.com/HegemanLab/w4mkmeans_galaxy_wrapper/tree/master commit 299e5c7fdb0d6eb0773f3660009f6d63c2082a8d
eschen42
parents:
diff changeset
210 nonzero.columns <- column.vars[nonzero.column.vars]
6ccbe18131a6 planemo upload for repository https://github.com/HegemanLab/w4mkmeans_galaxy_wrapper/tree/master commit 299e5c7fdb0d6eb0773f3660009f6d63c2082a8d
eschen42
parents:
diff changeset
211 if ( length(colnames(x)) != length(colnames(nonzero.columns)) ) {
6ccbe18131a6 planemo upload for repository https://github.com/HegemanLab/w4mkmeans_galaxy_wrapper/tree/master commit 299e5c7fdb0d6eb0773f3660009f6d63c2082a8d
eschen42
parents:
diff changeset
212 column.names <- attr(nonzero.columns,"names")
6ccbe18131a6 planemo upload for repository https://github.com/HegemanLab/w4mkmeans_galaxy_wrapper/tree/master commit 299e5c7fdb0d6eb0773f3660009f6d63c2082a8d
eschen42
parents:
diff changeset
213 x <- x[ , column.names, drop = FALSE ]
6ccbe18131a6 planemo upload for repository https://github.com/HegemanLab/w4mkmeans_galaxy_wrapper/tree/master commit 299e5c7fdb0d6eb0773f3660009f6d63c2082a8d
eschen42
parents:
diff changeset
214 }
6ccbe18131a6 planemo upload for repository https://github.com/HegemanLab/w4mkmeans_galaxy_wrapper/tree/master commit 299e5c7fdb0d6eb0773f3660009f6d63c2082a8d
eschen42
parents:
diff changeset
215 }
6ccbe18131a6 planemo upload for repository https://github.com/HegemanLab/w4mkmeans_galaxy_wrapper/tree/master commit 299e5c7fdb0d6eb0773f3660009f6d63c2082a8d
eschen42
parents:
diff changeset
216 return (x)
6ccbe18131a6 planemo upload for repository https://github.com/HegemanLab/w4mkmeans_galaxy_wrapper/tree/master commit 299e5c7fdb0d6eb0773f3660009f6d63c2082a8d
eschen42
parents:
diff changeset
217 }
6ccbe18131a6 planemo upload for repository https://github.com/HegemanLab/w4mkmeans_galaxy_wrapper/tree/master commit 299e5c7fdb0d6eb0773f3660009f6d63c2082a8d
eschen42
parents:
diff changeset
218
6ccbe18131a6 planemo upload for repository https://github.com/HegemanLab/w4mkmeans_galaxy_wrapper/tree/master commit 299e5c7fdb0d6eb0773f3660009f6d63c2082a8d
eschen42
parents:
diff changeset
219 if (is.null(x.matrix)) {
6ccbe18131a6 planemo upload for repository https://github.com/HegemanLab/w4mkmeans_galaxy_wrapper/tree/master commit 299e5c7fdb0d6eb0773f3660009f6d63c2082a8d
eschen42
parents:
diff changeset
220 stop("FATAL ERROR - prepare.data.matrix was called with null x.matrix")
6ccbe18131a6 planemo upload for repository https://github.com/HegemanLab/w4mkmeans_galaxy_wrapper/tree/master commit 299e5c7fdb0d6eb0773f3660009f6d63c2082a8d
eschen42
parents:
diff changeset
221 }
6ccbe18131a6 planemo upload for repository https://github.com/HegemanLab/w4mkmeans_galaxy_wrapper/tree/master commit 299e5c7fdb0d6eb0773f3660009f6d63c2082a8d
eschen42
parents:
diff changeset
222
2
c415b7dc6f37 planemo upload for repository https://github.com/HegemanLab/w4mkmeans_galaxy_wrapper/tree/master commit 3e916537da6bb37e6f3927d7a11e98e0ab6ef5ec
eschen42
parents: 0
diff changeset
223 enlog("prepare.data.matrix - get matrix")
c415b7dc6f37 planemo upload for repository https://github.com/HegemanLab/w4mkmeans_galaxy_wrapper/tree/master commit 3e916537da6bb37e6f3927d7a11e98e0ab6ef5ec
eschen42
parents: 0
diff changeset
224
0
6ccbe18131a6 planemo upload for repository https://github.com/HegemanLab/w4mkmeans_galaxy_wrapper/tree/master commit 299e5c7fdb0d6eb0773f3660009f6d63c2082a8d
eschen42
parents:
diff changeset
225 en$xpre <- x <- x.matrix
6ccbe18131a6 planemo upload for repository https://github.com/HegemanLab/w4mkmeans_galaxy_wrapper/tree/master commit 299e5c7fdb0d6eb0773f3660009f6d63c2082a8d
eschen42
parents:
diff changeset
226
6ccbe18131a6 planemo upload for repository https://github.com/HegemanLab/w4mkmeans_galaxy_wrapper/tree/master commit 299e5c7fdb0d6eb0773f3660009f6d63c2082a8d
eschen42
parents:
diff changeset
227 # exclude any samples as indicated
6ccbe18131a6 planemo upload for repository https://github.com/HegemanLab/w4mkmeans_galaxy_wrapper/tree/master commit 299e5c7fdb0d6eb0773f3660009f6d63c2082a8d
eschen42
parents:
diff changeset
228 if ( !is.null(exclude.features) ) {
2
c415b7dc6f37 planemo upload for repository https://github.com/HegemanLab/w4mkmeans_galaxy_wrapper/tree/master commit 3e916537da6bb37e6f3927d7a11e98e0ab6ef5ec
eschen42
parents: 0
diff changeset
229 enlog("prepare.data.matrix - exclude any samples as indicated")
0
6ccbe18131a6 planemo upload for repository https://github.com/HegemanLab/w4mkmeans_galaxy_wrapper/tree/master commit 299e5c7fdb0d6eb0773f3660009f6d63c2082a8d
eschen42
parents:
diff changeset
230 my.colnames <- colnames(x)
6ccbe18131a6 planemo upload for repository https://github.com/HegemanLab/w4mkmeans_galaxy_wrapper/tree/master commit 299e5c7fdb0d6eb0773f3660009f6d63c2082a8d
eschen42
parents:
diff changeset
231 my.col.diff <- setdiff(my.colnames, exclude.features)
6ccbe18131a6 planemo upload for repository https://github.com/HegemanLab/w4mkmeans_galaxy_wrapper/tree/master commit 299e5c7fdb0d6eb0773f3660009f6d63c2082a8d
eschen42
parents:
diff changeset
232 x <- x[ , my.col.diff , drop = FALSE ]
6ccbe18131a6 planemo upload for repository https://github.com/HegemanLab/w4mkmeans_galaxy_wrapper/tree/master commit 299e5c7fdb0d6eb0773f3660009f6d63c2082a8d
eschen42
parents:
diff changeset
233 }
6ccbe18131a6 planemo upload for repository https://github.com/HegemanLab/w4mkmeans_galaxy_wrapper/tree/master commit 299e5c7fdb0d6eb0773f3660009f6d63c2082a8d
eschen42
parents:
diff changeset
234
6ccbe18131a6 planemo upload for repository https://github.com/HegemanLab/w4mkmeans_galaxy_wrapper/tree/master commit 299e5c7fdb0d6eb0773f3660009f6d63c2082a8d
eschen42
parents:
diff changeset
235 # exclude any features as indicated
6ccbe18131a6 planemo upload for repository https://github.com/HegemanLab/w4mkmeans_galaxy_wrapper/tree/master commit 299e5c7fdb0d6eb0773f3660009f6d63c2082a8d
eschen42
parents:
diff changeset
236 if ( !is.null(exclude.samples) ) {
2
c415b7dc6f37 planemo upload for repository https://github.com/HegemanLab/w4mkmeans_galaxy_wrapper/tree/master commit 3e916537da6bb37e6f3927d7a11e98e0ab6ef5ec
eschen42
parents: 0
diff changeset
237 enlog("prepare.data.matrix - exclude any features as indicated")
0
6ccbe18131a6 planemo upload for repository https://github.com/HegemanLab/w4mkmeans_galaxy_wrapper/tree/master commit 299e5c7fdb0d6eb0773f3660009f6d63c2082a8d
eschen42
parents:
diff changeset
238 my.rownames <- rownames(x)
6ccbe18131a6 planemo upload for repository https://github.com/HegemanLab/w4mkmeans_galaxy_wrapper/tree/master commit 299e5c7fdb0d6eb0773f3660009f6d63c2082a8d
eschen42
parents:
diff changeset
239 my.row.diff <- setdiff(my.rownames, exclude.samples)
6ccbe18131a6 planemo upload for repository https://github.com/HegemanLab/w4mkmeans_galaxy_wrapper/tree/master commit 299e5c7fdb0d6eb0773f3660009f6d63c2082a8d
eschen42
parents:
diff changeset
240 x <- x[ my.row.diff, , drop = FALSE ]
6ccbe18131a6 planemo upload for repository https://github.com/HegemanLab/w4mkmeans_galaxy_wrapper/tree/master commit 299e5c7fdb0d6eb0773f3660009f6d63c2082a8d
eschen42
parents:
diff changeset
241 }
6ccbe18131a6 planemo upload for repository https://github.com/HegemanLab/w4mkmeans_galaxy_wrapper/tree/master commit 299e5c7fdb0d6eb0773f3660009f6d63c2082a8d
eschen42
parents:
diff changeset
242
6ccbe18131a6 planemo upload for repository https://github.com/HegemanLab/w4mkmeans_galaxy_wrapper/tree/master commit 299e5c7fdb0d6eb0773f3660009f6d63c2082a8d
eschen42
parents:
diff changeset
243 # rename rows if desired
6ccbe18131a6 planemo upload for repository https://github.com/HegemanLab/w4mkmeans_galaxy_wrapper/tree/master commit 299e5c7fdb0d6eb0773f3660009f6d63c2082a8d
eschen42
parents:
diff changeset
244 if ( !is.null(sample.rename.function) ) {
2
c415b7dc6f37 planemo upload for repository https://github.com/HegemanLab/w4mkmeans_galaxy_wrapper/tree/master commit 3e916537da6bb37e6f3927d7a11e98e0ab6ef5ec
eschen42
parents: 0
diff changeset
245 enlog("prepare.data.matrix - rename rows if desired")
0
6ccbe18131a6 planemo upload for repository https://github.com/HegemanLab/w4mkmeans_galaxy_wrapper/tree/master commit 299e5c7fdb0d6eb0773f3660009f6d63c2082a8d
eschen42
parents:
diff changeset
246 renamed <- sample.rename.function(x)
6ccbe18131a6 planemo upload for repository https://github.com/HegemanLab/w4mkmeans_galaxy_wrapper/tree/master commit 299e5c7fdb0d6eb0773f3660009f6d63c2082a8d
eschen42
parents:
diff changeset
247 rownames(x) <- renamed
6ccbe18131a6 planemo upload for repository https://github.com/HegemanLab/w4mkmeans_galaxy_wrapper/tree/master commit 299e5c7fdb0d6eb0773f3660009f6d63c2082a8d
eschen42
parents:
diff changeset
248 }
6ccbe18131a6 planemo upload for repository https://github.com/HegemanLab/w4mkmeans_galaxy_wrapper/tree/master commit 299e5c7fdb0d6eb0773f3660009f6d63c2082a8d
eschen42
parents:
diff changeset
249
2
c415b7dc6f37 planemo upload for repository https://github.com/HegemanLab/w4mkmeans_galaxy_wrapper/tree/master commit 3e916537da6bb37e6f3927d7a11e98e0ab6ef5ec
eschen42
parents: 0
diff changeset
250 enlog("prepare.data.matrix - save redacted x.datamatrix to environment")
c415b7dc6f37 planemo upload for repository https://github.com/HegemanLab/w4mkmeans_galaxy_wrapper/tree/master commit 3e916537da6bb37e6f3927d7a11e98e0ab6ef5ec
eschen42
parents: 0
diff changeset
251
0
6ccbe18131a6 planemo upload for repository https://github.com/HegemanLab/w4mkmeans_galaxy_wrapper/tree/master commit 299e5c7fdb0d6eb0773f3660009f6d63c2082a8d
eschen42
parents:
diff changeset
252 # save redacted x.datamatrix to environment
6ccbe18131a6 planemo upload for repository https://github.com/HegemanLab/w4mkmeans_galaxy_wrapper/tree/master commit 299e5c7fdb0d6eb0773f3660009f6d63c2082a8d
eschen42
parents:
diff changeset
253 en$redacted.data.matrix <- x
6ccbe18131a6 planemo upload for repository https://github.com/HegemanLab/w4mkmeans_galaxy_wrapper/tree/master commit 299e5c7fdb0d6eb0773f3660009f6d63c2082a8d
eschen42
parents:
diff changeset
254
6ccbe18131a6 planemo upload for repository https://github.com/HegemanLab/w4mkmeans_galaxy_wrapper/tree/master commit 299e5c7fdb0d6eb0773f3660009f6d63c2082a8d
eschen42
parents:
diff changeset
255 # impute values missing from the x.datamatrix
6ccbe18131a6 planemo upload for repository https://github.com/HegemanLab/w4mkmeans_galaxy_wrapper/tree/master commit 299e5c7fdb0d6eb0773f3660009f6d63c2082a8d
eschen42
parents:
diff changeset
256 if ( !is.null(data.imputation) ) {
2
c415b7dc6f37 planemo upload for repository https://github.com/HegemanLab/w4mkmeans_galaxy_wrapper/tree/master commit 3e916537da6bb37e6f3927d7a11e98e0ab6ef5ec
eschen42
parents: 0
diff changeset
257 enlog("prepare.data.matrix - impute values missing from the x.datamatrix")
0
6ccbe18131a6 planemo upload for repository https://github.com/HegemanLab/w4mkmeans_galaxy_wrapper/tree/master commit 299e5c7fdb0d6eb0773f3660009f6d63c2082a8d
eschen42
parents:
diff changeset
258 x <- data.imputation(x)
6ccbe18131a6 planemo upload for repository https://github.com/HegemanLab/w4mkmeans_galaxy_wrapper/tree/master commit 299e5c7fdb0d6eb0773f3660009f6d63c2082a8d
eschen42
parents:
diff changeset
259 }
6ccbe18131a6 planemo upload for repository https://github.com/HegemanLab/w4mkmeans_galaxy_wrapper/tree/master commit 299e5c7fdb0d6eb0773f3660009f6d63c2082a8d
eschen42
parents:
diff changeset
260
6ccbe18131a6 planemo upload for repository https://github.com/HegemanLab/w4mkmeans_galaxy_wrapper/tree/master commit 299e5c7fdb0d6eb0773f3660009f6d63c2082a8d
eschen42
parents:
diff changeset
261 # perform transformation if desired
6ccbe18131a6 planemo upload for repository https://github.com/HegemanLab/w4mkmeans_galaxy_wrapper/tree/master commit 299e5c7fdb0d6eb0773f3660009f6d63c2082a8d
eschen42
parents:
diff changeset
262 if ( !is.null(data.transformation) ) {
2
c415b7dc6f37 planemo upload for repository https://github.com/HegemanLab/w4mkmeans_galaxy_wrapper/tree/master commit 3e916537da6bb37e6f3927d7a11e98e0ab6ef5ec
eschen42
parents: 0
diff changeset
263 enlog("prepare.data.matrix - perform transformation")
0
6ccbe18131a6 planemo upload for repository https://github.com/HegemanLab/w4mkmeans_galaxy_wrapper/tree/master commit 299e5c7fdb0d6eb0773f3660009f6d63c2082a8d
eschen42
parents:
diff changeset
264 x <- data.transformation(x)
6ccbe18131a6 planemo upload for repository https://github.com/HegemanLab/w4mkmeans_galaxy_wrapper/tree/master commit 299e5c7fdb0d6eb0773f3660009f6d63c2082a8d
eschen42
parents:
diff changeset
265 } else {
6ccbe18131a6 planemo upload for repository https://github.com/HegemanLab/w4mkmeans_galaxy_wrapper/tree/master commit 299e5c7fdb0d6eb0773f3660009f6d63c2082a8d
eschen42
parents:
diff changeset
266 x <- x
6ccbe18131a6 planemo upload for repository https://github.com/HegemanLab/w4mkmeans_galaxy_wrapper/tree/master commit 299e5c7fdb0d6eb0773f3660009f6d63c2082a8d
eschen42
parents:
diff changeset
267 }
6ccbe18131a6 planemo upload for repository https://github.com/HegemanLab/w4mkmeans_galaxy_wrapper/tree/master commit 299e5c7fdb0d6eb0773f3660009f6d63c2082a8d
eschen42
parents:
diff changeset
268
6ccbe18131a6 planemo upload for repository https://github.com/HegemanLab/w4mkmeans_galaxy_wrapper/tree/master commit 299e5c7fdb0d6eb0773f3660009f6d63c2082a8d
eschen42
parents:
diff changeset
269 # purge rows and columns that have zero variance
6ccbe18131a6 planemo upload for repository https://github.com/HegemanLab/w4mkmeans_galaxy_wrapper/tree/master commit 299e5c7fdb0d6eb0773f3660009f6d63c2082a8d
eschen42
parents:
diff changeset
270 if ( is.numeric(x) ) {
2
c415b7dc6f37 planemo upload for repository https://github.com/HegemanLab/w4mkmeans_galaxy_wrapper/tree/master commit 3e916537da6bb37e6f3927d7a11e98e0ab6ef5ec
eschen42
parents: 0
diff changeset
271 enlog("prepare.data.matrix - purge rows and columns that have zero variance")
0
6ccbe18131a6 planemo upload for repository https://github.com/HegemanLab/w4mkmeans_galaxy_wrapper/tree/master commit 299e5c7fdb0d6eb0773f3660009f6d63c2082a8d
eschen42
parents:
diff changeset
272 x <- nonzero.var(x)
6ccbe18131a6 planemo upload for repository https://github.com/HegemanLab/w4mkmeans_galaxy_wrapper/tree/master commit 299e5c7fdb0d6eb0773f3660009f6d63c2082a8d
eschen42
parents:
diff changeset
273 }
6ccbe18131a6 planemo upload for repository https://github.com/HegemanLab/w4mkmeans_galaxy_wrapper/tree/master commit 299e5c7fdb0d6eb0773f3660009f6d63c2082a8d
eschen42
parents:
diff changeset
274
6ccbe18131a6 planemo upload for repository https://github.com/HegemanLab/w4mkmeans_galaxy_wrapper/tree/master commit 299e5c7fdb0d6eb0773f3660009f6d63c2082a8d
eschen42
parents:
diff changeset
275 # save imputed, transformed x.datamatrix to environment
6ccbe18131a6 planemo upload for repository https://github.com/HegemanLab/w4mkmeans_galaxy_wrapper/tree/master commit 299e5c7fdb0d6eb0773f3660009f6d63c2082a8d
eschen42
parents:
diff changeset
276 en$imputed.transformed.data.matrix <- x
6ccbe18131a6 planemo upload for repository https://github.com/HegemanLab/w4mkmeans_galaxy_wrapper/tree/master commit 299e5c7fdb0d6eb0773f3660009f6d63c2082a8d
eschen42
parents:
diff changeset
277
6ccbe18131a6 planemo upload for repository https://github.com/HegemanLab/w4mkmeans_galaxy_wrapper/tree/master commit 299e5c7fdb0d6eb0773f3660009f6d63c2082a8d
eschen42
parents:
diff changeset
278 return(x)
6ccbe18131a6 planemo upload for repository https://github.com/HegemanLab/w4mkmeans_galaxy_wrapper/tree/master commit 299e5c7fdb0d6eb0773f3660009f6d63c2082a8d
eschen42
parents:
diff changeset
279 }
6ccbe18131a6 planemo upload for repository https://github.com/HegemanLab/w4mkmeans_galaxy_wrapper/tree/master commit 299e5c7fdb0d6eb0773f3660009f6d63c2082a8d
eschen42
parents:
diff changeset
280
2
c415b7dc6f37 planemo upload for repository https://github.com/HegemanLab/w4mkmeans_galaxy_wrapper/tree/master commit 3e916537da6bb37e6f3927d7a11e98e0ab6ef5ec
eschen42
parents: 0
diff changeset
281 # vim: sw=2 ts=2 et :