annotate w4mcorcov_salience.R @ 7:066b1f409e9f draft

planemo upload for repository https://github.com/HegemanLab/w4mcorcov_galaxy_wrapper/tree/master commit e73fabe1b3c871abbcb2e89914c181149c8e2066
author eschen42
date Thu, 26 Jul 2018 10:23:34 -0400
parents 23f9fad4edfc
children 06c51af11531
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
0
23f9fad4edfc planemo upload for repository https://github.com/HegemanLab/w4mcorcov_galaxy_wrapper/tree/master commit bd26542b811de06c1a877337a2840a9f899c2b94
eschen42
parents:
diff changeset
1 w4msalience <- function(
23f9fad4edfc planemo upload for repository https://github.com/HegemanLab/w4mcorcov_galaxy_wrapper/tree/master commit bd26542b811de06c1a877337a2840a9f899c2b94
eschen42
parents:
diff changeset
2 data_matrix # a matrix of intensities; features as rows, and samples as columns
23f9fad4edfc planemo upload for repository https://github.com/HegemanLab/w4mcorcov_galaxy_wrapper/tree/master commit bd26542b811de06c1a877337a2840a9f899c2b94
eschen42
parents:
diff changeset
3 , sample_class # a vector of sample class-levels; length(sample_class) == ncol(data_matrix)
23f9fad4edfc planemo upload for repository https://github.com/HegemanLab/w4mcorcov_galaxy_wrapper/tree/master commit bd26542b811de06c1a877337a2840a9f899c2b94
eschen42
parents:
diff changeset
4 , failure_action = stop
23f9fad4edfc planemo upload for repository https://github.com/HegemanLab/w4mcorcov_galaxy_wrapper/tree/master commit bd26542b811de06c1a877337a2840a9f899c2b94
eschen42
parents:
diff changeset
5 ) {
23f9fad4edfc planemo upload for repository https://github.com/HegemanLab/w4mcorcov_galaxy_wrapper/tree/master commit bd26542b811de06c1a877337a2840a9f899c2b94
eschen42
parents:
diff changeset
6 library(stats)
23f9fad4edfc planemo upload for repository https://github.com/HegemanLab/w4mcorcov_galaxy_wrapper/tree/master commit bd26542b811de06c1a877337a2840a9f899c2b94
eschen42
parents:
diff changeset
7 # begin sanity checks
23f9fad4edfc planemo upload for repository https://github.com/HegemanLab/w4mcorcov_galaxy_wrapper/tree/master commit bd26542b811de06c1a877337a2840a9f899c2b94
eschen42
parents:
diff changeset
8 if ( !is.vector(sample_class) || !( is.character(sample_class) || is.factor(sample_class) ) ) {
23f9fad4edfc planemo upload for repository https://github.com/HegemanLab/w4mcorcov_galaxy_wrapper/tree/master commit bd26542b811de06c1a877337a2840a9f899c2b94
eschen42
parents:
diff changeset
9 failure_action("w4msalience: Expected sample_class to be a vector of characters of factor-levels")
23f9fad4edfc planemo upload for repository https://github.com/HegemanLab/w4mcorcov_galaxy_wrapper/tree/master commit bd26542b811de06c1a877337a2840a9f899c2b94
eschen42
parents:
diff changeset
10 return (NULL)
23f9fad4edfc planemo upload for repository https://github.com/HegemanLab/w4mcorcov_galaxy_wrapper/tree/master commit bd26542b811de06c1a877337a2840a9f899c2b94
eschen42
parents:
diff changeset
11 }
23f9fad4edfc planemo upload for repository https://github.com/HegemanLab/w4mcorcov_galaxy_wrapper/tree/master commit bd26542b811de06c1a877337a2840a9f899c2b94
eschen42
parents:
diff changeset
12 if ( !is.matrix(data_matrix) && !is.data.frame(data_matrix) ) {
23f9fad4edfc planemo upload for repository https://github.com/HegemanLab/w4mcorcov_galaxy_wrapper/tree/master commit bd26542b811de06c1a877337a2840a9f899c2b94
eschen42
parents:
diff changeset
13 failure_action("w4msalience: Expected data_matrix to be a matrix (or data.frame) of numeric")
23f9fad4edfc planemo upload for repository https://github.com/HegemanLab/w4mcorcov_galaxy_wrapper/tree/master commit bd26542b811de06c1a877337a2840a9f899c2b94
eschen42
parents:
diff changeset
14 return (NULL)
23f9fad4edfc planemo upload for repository https://github.com/HegemanLab/w4mcorcov_galaxy_wrapper/tree/master commit bd26542b811de06c1a877337a2840a9f899c2b94
eschen42
parents:
diff changeset
15 }
23f9fad4edfc planemo upload for repository https://github.com/HegemanLab/w4mcorcov_galaxy_wrapper/tree/master commit bd26542b811de06c1a877337a2840a9f899c2b94
eschen42
parents:
diff changeset
16 # transpose data_matrix so that columns are the features
23f9fad4edfc planemo upload for repository https://github.com/HegemanLab/w4mcorcov_galaxy_wrapper/tree/master commit bd26542b811de06c1a877337a2840a9f899c2b94
eschen42
parents:
diff changeset
17 t_data_matrix <- t(data_matrix)
23f9fad4edfc planemo upload for repository https://github.com/HegemanLab/w4mcorcov_galaxy_wrapper/tree/master commit bd26542b811de06c1a877337a2840a9f899c2b94
eschen42
parents:
diff changeset
18 if ( !is.matrix(t_data_matrix) || !is.numeric(t_data_matrix) ) {
23f9fad4edfc planemo upload for repository https://github.com/HegemanLab/w4mcorcov_galaxy_wrapper/tree/master commit bd26542b811de06c1a877337a2840a9f899c2b94
eschen42
parents:
diff changeset
19 failure_action("w4msalience: Expected data_matrix to be a matrix (or data.frame) of numeric")
23f9fad4edfc planemo upload for repository https://github.com/HegemanLab/w4mcorcov_galaxy_wrapper/tree/master commit bd26542b811de06c1a877337a2840a9f899c2b94
eschen42
parents:
diff changeset
20 return (NULL)
23f9fad4edfc planemo upload for repository https://github.com/HegemanLab/w4mcorcov_galaxy_wrapper/tree/master commit bd26542b811de06c1a877337a2840a9f899c2b94
eschen42
parents:
diff changeset
21 }
23f9fad4edfc planemo upload for repository https://github.com/HegemanLab/w4mcorcov_galaxy_wrapper/tree/master commit bd26542b811de06c1a877337a2840a9f899c2b94
eschen42
parents:
diff changeset
22 n_features <- ncol(t_data_matrix)
23f9fad4edfc planemo upload for repository https://github.com/HegemanLab/w4mcorcov_galaxy_wrapper/tree/master commit bd26542b811de06c1a877337a2840a9f899c2b94
eschen42
parents:
diff changeset
23 n_features_plus_1 <- 1 + n_features
23f9fad4edfc planemo upload for repository https://github.com/HegemanLab/w4mcorcov_galaxy_wrapper/tree/master commit bd26542b811de06c1a877337a2840a9f899c2b94
eschen42
parents:
diff changeset
24 features <- colnames(t_data_matrix)
23f9fad4edfc planemo upload for repository https://github.com/HegemanLab/w4mcorcov_galaxy_wrapper/tree/master commit bd26542b811de06c1a877337a2840a9f899c2b94
eschen42
parents:
diff changeset
25 n_samples <- nrow(t_data_matrix)
23f9fad4edfc planemo upload for repository https://github.com/HegemanLab/w4mcorcov_galaxy_wrapper/tree/master commit bd26542b811de06c1a877337a2840a9f899c2b94
eschen42
parents:
diff changeset
26 if ( length(sample_class) != n_samples ) {
23f9fad4edfc planemo upload for repository https://github.com/HegemanLab/w4mcorcov_galaxy_wrapper/tree/master commit bd26542b811de06c1a877337a2840a9f899c2b94
eschen42
parents:
diff changeset
27 strF(data_matrix)
23f9fad4edfc planemo upload for repository https://github.com/HegemanLab/w4mcorcov_galaxy_wrapper/tree/master commit bd26542b811de06c1a877337a2840a9f899c2b94
eschen42
parents:
diff changeset
28 strF(sample_class)
23f9fad4edfc planemo upload for repository https://github.com/HegemanLab/w4mcorcov_galaxy_wrapper/tree/master commit bd26542b811de06c1a877337a2840a9f899c2b94
eschen42
parents:
diff changeset
29 failure_action(sprintf("w4msalience: The data_matrix has %d samples but sample_class has %d", n_samples, length(sample_class)))
23f9fad4edfc planemo upload for repository https://github.com/HegemanLab/w4mcorcov_galaxy_wrapper/tree/master commit bd26542b811de06c1a877337a2840a9f899c2b94
eschen42
parents:
diff changeset
30 return (NULL)
23f9fad4edfc planemo upload for repository https://github.com/HegemanLab/w4mcorcov_galaxy_wrapper/tree/master commit bd26542b811de06c1a877337a2840a9f899c2b94
eschen42
parents:
diff changeset
31 }
23f9fad4edfc planemo upload for repository https://github.com/HegemanLab/w4mcorcov_galaxy_wrapper/tree/master commit bd26542b811de06c1a877337a2840a9f899c2b94
eschen42
parents:
diff changeset
32 # end sanity checks
23f9fad4edfc planemo upload for repository https://github.com/HegemanLab/w4mcorcov_galaxy_wrapper/tree/master commit bd26542b811de06c1a877337a2840a9f899c2b94
eschen42
parents:
diff changeset
33
23f9fad4edfc planemo upload for repository https://github.com/HegemanLab/w4mcorcov_galaxy_wrapper/tree/master commit bd26542b811de06c1a877337a2840a9f899c2b94
eschen42
parents:
diff changeset
34 # "For each feature, 'select sample_class, median(intensity) from feature group by sample_class'."
23f9fad4edfc planemo upload for repository https://github.com/HegemanLab/w4mcorcov_galaxy_wrapper/tree/master commit bd26542b811de06c1a877337a2840a9f899c2b94
eschen42
parents:
diff changeset
35 # The first column(s) of the result of aggregate has the classifier value(s) specified in the 'by' list.
23f9fad4edfc planemo upload for repository https://github.com/HegemanLab/w4mcorcov_galaxy_wrapper/tree/master commit bd26542b811de06c1a877337a2840a9f899c2b94
eschen42
parents:
diff changeset
36 medianOfFeatureBySampleClassLevel <- aggregate(
23f9fad4edfc planemo upload for repository https://github.com/HegemanLab/w4mcorcov_galaxy_wrapper/tree/master commit bd26542b811de06c1a877337a2840a9f899c2b94
eschen42
parents:
diff changeset
37 x = as.data.frame(t_data_matrix)
23f9fad4edfc planemo upload for repository https://github.com/HegemanLab/w4mcorcov_galaxy_wrapper/tree/master commit bd26542b811de06c1a877337a2840a9f899c2b94
eschen42
parents:
diff changeset
38 , by = list(sample_class)
23f9fad4edfc planemo upload for repository https://github.com/HegemanLab/w4mcorcov_galaxy_wrapper/tree/master commit bd26542b811de06c1a877337a2840a9f899c2b94
eschen42
parents:
diff changeset
39 , FUN = "median"
23f9fad4edfc planemo upload for repository https://github.com/HegemanLab/w4mcorcov_galaxy_wrapper/tree/master commit bd26542b811de06c1a877337a2840a9f899c2b94
eschen42
parents:
diff changeset
40 )
23f9fad4edfc planemo upload for repository https://github.com/HegemanLab/w4mcorcov_galaxy_wrapper/tree/master commit bd26542b811de06c1a877337a2840a9f899c2b94
eschen42
parents:
diff changeset
41
23f9fad4edfc planemo upload for repository https://github.com/HegemanLab/w4mcorcov_galaxy_wrapper/tree/master commit bd26542b811de06c1a877337a2840a9f899c2b94
eschen42
parents:
diff changeset
42 # "For each feature, 'select sample_class, max(intensity) from feature group by sample_class'."
23f9fad4edfc planemo upload for repository https://github.com/HegemanLab/w4mcorcov_galaxy_wrapper/tree/master commit bd26542b811de06c1a877337a2840a9f899c2b94
eschen42
parents:
diff changeset
43 maxOfFeatureBySampleClassLevel <- aggregate(
23f9fad4edfc planemo upload for repository https://github.com/HegemanLab/w4mcorcov_galaxy_wrapper/tree/master commit bd26542b811de06c1a877337a2840a9f899c2b94
eschen42
parents:
diff changeset
44 x = as.data.frame(t_data_matrix)
23f9fad4edfc planemo upload for repository https://github.com/HegemanLab/w4mcorcov_galaxy_wrapper/tree/master commit bd26542b811de06c1a877337a2840a9f899c2b94
eschen42
parents:
diff changeset
45 , by = list(sample_class)
23f9fad4edfc planemo upload for repository https://github.com/HegemanLab/w4mcorcov_galaxy_wrapper/tree/master commit bd26542b811de06c1a877337a2840a9f899c2b94
eschen42
parents:
diff changeset
46 , FUN = "max"
23f9fad4edfc planemo upload for repository https://github.com/HegemanLab/w4mcorcov_galaxy_wrapper/tree/master commit bd26542b811de06c1a877337a2840a9f899c2b94
eschen42
parents:
diff changeset
47 )
23f9fad4edfc planemo upload for repository https://github.com/HegemanLab/w4mcorcov_galaxy_wrapper/tree/master commit bd26542b811de06c1a877337a2840a9f899c2b94
eschen42
parents:
diff changeset
48
23f9fad4edfc planemo upload for repository https://github.com/HegemanLab/w4mcorcov_galaxy_wrapper/tree/master commit bd26542b811de06c1a877337a2840a9f899c2b94
eschen42
parents:
diff changeset
49 # "For each feature, 'select sample_class, rcv(intensity) from feature group by sample_class'."
23f9fad4edfc planemo upload for repository https://github.com/HegemanLab/w4mcorcov_galaxy_wrapper/tree/master commit bd26542b811de06c1a877337a2840a9f899c2b94
eschen42
parents:
diff changeset
50 # cv is less robust; deviation from normality degrades performance
23f9fad4edfc planemo upload for repository https://github.com/HegemanLab/w4mcorcov_galaxy_wrapper/tree/master commit bd26542b811de06c1a877337a2840a9f899c2b94
eschen42
parents:
diff changeset
51 # cv(x) == sd(x) / mean(x)
23f9fad4edfc planemo upload for repository https://github.com/HegemanLab/w4mcorcov_galaxy_wrapper/tree/master commit bd26542b811de06c1a877337a2840a9f899c2b94
eschen42
parents:
diff changeset
52 # rcv is a "robust" coefficient of variation, expressed as a proportion
23f9fad4edfc planemo upload for repository https://github.com/HegemanLab/w4mcorcov_galaxy_wrapper/tree/master commit bd26542b811de06c1a877337a2840a9f899c2b94
eschen42
parents:
diff changeset
53 # rcv(x) == mad(x) / median(x)
23f9fad4edfc planemo upload for repository https://github.com/HegemanLab/w4mcorcov_galaxy_wrapper/tree/master commit bd26542b811de06c1a877337a2840a9f899c2b94
eschen42
parents:
diff changeset
54 madOfFeatureBySampleClassLevel <- aggregate(
23f9fad4edfc planemo upload for repository https://github.com/HegemanLab/w4mcorcov_galaxy_wrapper/tree/master commit bd26542b811de06c1a877337a2840a9f899c2b94
eschen42
parents:
diff changeset
55 x = as.data.frame(t_data_matrix)
23f9fad4edfc planemo upload for repository https://github.com/HegemanLab/w4mcorcov_galaxy_wrapper/tree/master commit bd26542b811de06c1a877337a2840a9f899c2b94
eschen42
parents:
diff changeset
56 , by = list(sample_class)
23f9fad4edfc planemo upload for repository https://github.com/HegemanLab/w4mcorcov_galaxy_wrapper/tree/master commit bd26542b811de06c1a877337a2840a9f899c2b94
eschen42
parents:
diff changeset
57 , FUN = "mad"
23f9fad4edfc planemo upload for repository https://github.com/HegemanLab/w4mcorcov_galaxy_wrapper/tree/master commit bd26542b811de06c1a877337a2840a9f899c2b94
eschen42
parents:
diff changeset
58 )
23f9fad4edfc planemo upload for repository https://github.com/HegemanLab/w4mcorcov_galaxy_wrapper/tree/master commit bd26542b811de06c1a877337a2840a9f899c2b94
eschen42
parents:
diff changeset
59 rcvOfFeatureBySampleClassLevel <- as.matrix(
23f9fad4edfc planemo upload for repository https://github.com/HegemanLab/w4mcorcov_galaxy_wrapper/tree/master commit bd26542b811de06c1a877337a2840a9f899c2b94
eschen42
parents:
diff changeset
60 madOfFeatureBySampleClassLevel[,2:n_features_plus_1] / medianOfFeatureBySampleClassLevel[,2:n_features_plus_1]
23f9fad4edfc planemo upload for repository https://github.com/HegemanLab/w4mcorcov_galaxy_wrapper/tree/master commit bd26542b811de06c1a877337a2840a9f899c2b94
eschen42
parents:
diff changeset
61 )
23f9fad4edfc planemo upload for repository https://github.com/HegemanLab/w4mcorcov_galaxy_wrapper/tree/master commit bd26542b811de06c1a877337a2840a9f899c2b94
eschen42
parents:
diff changeset
62 rcvOfFeatureBySampleClassLevel[is.nan(rcvOfFeatureBySampleClassLevel)] <- max(9999,max(rcvOfFeatureBySampleClassLevel, na.rm = TRUE))
23f9fad4edfc planemo upload for repository https://github.com/HegemanLab/w4mcorcov_galaxy_wrapper/tree/master commit bd26542b811de06c1a877337a2840a9f899c2b94
eschen42
parents:
diff changeset
63
23f9fad4edfc planemo upload for repository https://github.com/HegemanLab/w4mcorcov_galaxy_wrapper/tree/master commit bd26542b811de06c1a877337a2840a9f899c2b94
eschen42
parents:
diff changeset
64 # "For each feature, 'select max(median_feature_intensity) from feature'."
23f9fad4edfc planemo upload for repository https://github.com/HegemanLab/w4mcorcov_galaxy_wrapper/tree/master commit bd26542b811de06c1a877337a2840a9f899c2b94
eschen42
parents:
diff changeset
65 maxApplyMedianOfFeatureBySampleClassLevel <- sapply(
23f9fad4edfc planemo upload for repository https://github.com/HegemanLab/w4mcorcov_galaxy_wrapper/tree/master commit bd26542b811de06c1a877337a2840a9f899c2b94
eschen42
parents:
diff changeset
66 X = 1:n_features
23f9fad4edfc planemo upload for repository https://github.com/HegemanLab/w4mcorcov_galaxy_wrapper/tree/master commit bd26542b811de06c1a877337a2840a9f899c2b94
eschen42
parents:
diff changeset
67 , FUN = function(i) {
23f9fad4edfc planemo upload for repository https://github.com/HegemanLab/w4mcorcov_galaxy_wrapper/tree/master commit bd26542b811de06c1a877337a2840a9f899c2b94
eschen42
parents:
diff changeset
68 match(
23f9fad4edfc planemo upload for repository https://github.com/HegemanLab/w4mcorcov_galaxy_wrapper/tree/master commit bd26542b811de06c1a877337a2840a9f899c2b94
eschen42
parents:
diff changeset
69 max(maxOfFeatureBySampleClassLevel[, i + 1])
23f9fad4edfc planemo upload for repository https://github.com/HegemanLab/w4mcorcov_galaxy_wrapper/tree/master commit bd26542b811de06c1a877337a2840a9f899c2b94
eschen42
parents:
diff changeset
70 , maxOfFeatureBySampleClassLevel[, i + 1]
23f9fad4edfc planemo upload for repository https://github.com/HegemanLab/w4mcorcov_galaxy_wrapper/tree/master commit bd26542b811de06c1a877337a2840a9f899c2b94
eschen42
parents:
diff changeset
71 )
23f9fad4edfc planemo upload for repository https://github.com/HegemanLab/w4mcorcov_galaxy_wrapper/tree/master commit bd26542b811de06c1a877337a2840a9f899c2b94
eschen42
parents:
diff changeset
72 }
23f9fad4edfc planemo upload for repository https://github.com/HegemanLab/w4mcorcov_galaxy_wrapper/tree/master commit bd26542b811de06c1a877337a2840a9f899c2b94
eschen42
parents:
diff changeset
73 )
23f9fad4edfc planemo upload for repository https://github.com/HegemanLab/w4mcorcov_galaxy_wrapper/tree/master commit bd26542b811de06c1a877337a2840a9f899c2b94
eschen42
parents:
diff changeset
74
23f9fad4edfc planemo upload for repository https://github.com/HegemanLab/w4mcorcov_galaxy_wrapper/tree/master commit bd26542b811de06c1a877337a2840a9f899c2b94
eschen42
parents:
diff changeset
75 # "For each feature, 'select mean(median_feature_intensity) from feature'."
23f9fad4edfc planemo upload for repository https://github.com/HegemanLab/w4mcorcov_galaxy_wrapper/tree/master commit bd26542b811de06c1a877337a2840a9f899c2b94
eschen42
parents:
diff changeset
76 meanApplyMedianOfFeatureBySampleClassLevel <- sapply(
23f9fad4edfc planemo upload for repository https://github.com/HegemanLab/w4mcorcov_galaxy_wrapper/tree/master commit bd26542b811de06c1a877337a2840a9f899c2b94
eschen42
parents:
diff changeset
77 X = 1:n_features
23f9fad4edfc planemo upload for repository https://github.com/HegemanLab/w4mcorcov_galaxy_wrapper/tree/master commit bd26542b811de06c1a877337a2840a9f899c2b94
eschen42
parents:
diff changeset
78 , FUN = function(i) mean(medianOfFeatureBySampleClassLevel[, i + 1])
23f9fad4edfc planemo upload for repository https://github.com/HegemanLab/w4mcorcov_galaxy_wrapper/tree/master commit bd26542b811de06c1a877337a2840a9f899c2b94
eschen42
parents:
diff changeset
79 )
23f9fad4edfc planemo upload for repository https://github.com/HegemanLab/w4mcorcov_galaxy_wrapper/tree/master commit bd26542b811de06c1a877337a2840a9f899c2b94
eschen42
parents:
diff changeset
80
23f9fad4edfc planemo upload for repository https://github.com/HegemanLab/w4mcorcov_galaxy_wrapper/tree/master commit bd26542b811de06c1a877337a2840a9f899c2b94
eschen42
parents:
diff changeset
81 # Compute the 'salience' for each feature, i.e., how salient the intensity of a feature
23f9fad4edfc planemo upload for repository https://github.com/HegemanLab/w4mcorcov_galaxy_wrapper/tree/master commit bd26542b811de06c1a877337a2840a9f899c2b94
eschen42
parents:
diff changeset
82 # is for one particular class-level relative to the intensity across all class-levels.
23f9fad4edfc planemo upload for repository https://github.com/HegemanLab/w4mcorcov_galaxy_wrapper/tree/master commit bd26542b811de06c1a877337a2840a9f899c2b94
eschen42
parents:
diff changeset
83 salience_df <- data.frame(
23f9fad4edfc planemo upload for repository https://github.com/HegemanLab/w4mcorcov_galaxy_wrapper/tree/master commit bd26542b811de06c1a877337a2840a9f899c2b94
eschen42
parents:
diff changeset
84 # the feature name
23f9fad4edfc planemo upload for repository https://github.com/HegemanLab/w4mcorcov_galaxy_wrapper/tree/master commit bd26542b811de06c1a877337a2840a9f899c2b94
eschen42
parents:
diff changeset
85 feature = features
23f9fad4edfc planemo upload for repository https://github.com/HegemanLab/w4mcorcov_galaxy_wrapper/tree/master commit bd26542b811de06c1a877337a2840a9f899c2b94
eschen42
parents:
diff changeset
86 # the name (or factor-level) of the class-level with the highest median intensity for the feature
23f9fad4edfc planemo upload for repository https://github.com/HegemanLab/w4mcorcov_galaxy_wrapper/tree/master commit bd26542b811de06c1a877337a2840a9f899c2b94
eschen42
parents:
diff changeset
87 , max_level = medianOfFeatureBySampleClassLevel[maxApplyMedianOfFeatureBySampleClassLevel,1]
23f9fad4edfc planemo upload for repository https://github.com/HegemanLab/w4mcorcov_galaxy_wrapper/tree/master commit bd26542b811de06c1a877337a2840a9f899c2b94
eschen42
parents:
diff changeset
88 # the median intensity for the feature and the level max_level
23f9fad4edfc planemo upload for repository https://github.com/HegemanLab/w4mcorcov_galaxy_wrapper/tree/master commit bd26542b811de06c1a877337a2840a9f899c2b94
eschen42
parents:
diff changeset
89 , max_median = sapply(
23f9fad4edfc planemo upload for repository https://github.com/HegemanLab/w4mcorcov_galaxy_wrapper/tree/master commit bd26542b811de06c1a877337a2840a9f899c2b94
eschen42
parents:
diff changeset
90 X = 1:n_features
23f9fad4edfc planemo upload for repository https://github.com/HegemanLab/w4mcorcov_galaxy_wrapper/tree/master commit bd26542b811de06c1a877337a2840a9f899c2b94
eschen42
parents:
diff changeset
91 , FUN = function(i) {
23f9fad4edfc planemo upload for repository https://github.com/HegemanLab/w4mcorcov_galaxy_wrapper/tree/master commit bd26542b811de06c1a877337a2840a9f899c2b94
eschen42
parents:
diff changeset
92 maxOfFeatureBySampleClassLevel[maxApplyMedianOfFeatureBySampleClassLevel[i], 1 + i]
23f9fad4edfc planemo upload for repository https://github.com/HegemanLab/w4mcorcov_galaxy_wrapper/tree/master commit bd26542b811de06c1a877337a2840a9f899c2b94
eschen42
parents:
diff changeset
93 }
23f9fad4edfc planemo upload for repository https://github.com/HegemanLab/w4mcorcov_galaxy_wrapper/tree/master commit bd26542b811de06c1a877337a2840a9f899c2b94
eschen42
parents:
diff changeset
94 )
23f9fad4edfc planemo upload for repository https://github.com/HegemanLab/w4mcorcov_galaxy_wrapper/tree/master commit bd26542b811de06c1a877337a2840a9f899c2b94
eschen42
parents:
diff changeset
95 # the coefficient of variation (expressed as a proportion) for the intensity for the feature and the level max_level
23f9fad4edfc planemo upload for repository https://github.com/HegemanLab/w4mcorcov_galaxy_wrapper/tree/master commit bd26542b811de06c1a877337a2840a9f899c2b94
eschen42
parents:
diff changeset
96 , max_rcv = sapply(
23f9fad4edfc planemo upload for repository https://github.com/HegemanLab/w4mcorcov_galaxy_wrapper/tree/master commit bd26542b811de06c1a877337a2840a9f899c2b94
eschen42
parents:
diff changeset
97 X = 1:n_features
23f9fad4edfc planemo upload for repository https://github.com/HegemanLab/w4mcorcov_galaxy_wrapper/tree/master commit bd26542b811de06c1a877337a2840a9f899c2b94
eschen42
parents:
diff changeset
98 , FUN = function(i) {
23f9fad4edfc planemo upload for repository https://github.com/HegemanLab/w4mcorcov_galaxy_wrapper/tree/master commit bd26542b811de06c1a877337a2840a9f899c2b94
eschen42
parents:
diff changeset
99 rcvOfFeatureBySampleClassLevel[maxApplyMedianOfFeatureBySampleClassLevel[i], i]
23f9fad4edfc planemo upload for repository https://github.com/HegemanLab/w4mcorcov_galaxy_wrapper/tree/master commit bd26542b811de06c1a877337a2840a9f899c2b94
eschen42
parents:
diff changeset
100 }
23f9fad4edfc planemo upload for repository https://github.com/HegemanLab/w4mcorcov_galaxy_wrapper/tree/master commit bd26542b811de06c1a877337a2840a9f899c2b94
eschen42
parents:
diff changeset
101 )
23f9fad4edfc planemo upload for repository https://github.com/HegemanLab/w4mcorcov_galaxy_wrapper/tree/master commit bd26542b811de06c1a877337a2840a9f899c2b94
eschen42
parents:
diff changeset
102 # the mean of the medians of intensity for all class-levels for the feature
23f9fad4edfc planemo upload for repository https://github.com/HegemanLab/w4mcorcov_galaxy_wrapper/tree/master commit bd26542b811de06c1a877337a2840a9f899c2b94
eschen42
parents:
diff changeset
103 , mean_median = meanApplyMedianOfFeatureBySampleClassLevel
23f9fad4edfc planemo upload for repository https://github.com/HegemanLab/w4mcorcov_galaxy_wrapper/tree/master commit bd26542b811de06c1a877337a2840a9f899c2b94
eschen42
parents:
diff changeset
104 # don't coerce strings to factors (this is a parameter for the data.frame constructor, not a column of the data.frame)
23f9fad4edfc planemo upload for repository https://github.com/HegemanLab/w4mcorcov_galaxy_wrapper/tree/master commit bd26542b811de06c1a877337a2840a9f899c2b94
eschen42
parents:
diff changeset
105 , stringsAsFactors = FALSE
23f9fad4edfc planemo upload for repository https://github.com/HegemanLab/w4mcorcov_galaxy_wrapper/tree/master commit bd26542b811de06c1a877337a2840a9f899c2b94
eschen42
parents:
diff changeset
106 )
23f9fad4edfc planemo upload for repository https://github.com/HegemanLab/w4mcorcov_galaxy_wrapper/tree/master commit bd26542b811de06c1a877337a2840a9f899c2b94
eschen42
parents:
diff changeset
107 # raw salience is the ratio of the most-prominent level to the mean of all levels for the feature
23f9fad4edfc planemo upload for repository https://github.com/HegemanLab/w4mcorcov_galaxy_wrapper/tree/master commit bd26542b811de06c1a877337a2840a9f899c2b94
eschen42
parents:
diff changeset
108 salience_df$salience <- sapply(
23f9fad4edfc planemo upload for repository https://github.com/HegemanLab/w4mcorcov_galaxy_wrapper/tree/master commit bd26542b811de06c1a877337a2840a9f899c2b94
eschen42
parents:
diff changeset
109 X = 1:nrow(salience_df)
23f9fad4edfc planemo upload for repository https://github.com/HegemanLab/w4mcorcov_galaxy_wrapper/tree/master commit bd26542b811de06c1a877337a2840a9f899c2b94
eschen42
parents:
diff changeset
110 , FUN = function(i) with(salience_df[i,], if (mean_median > 0) { max_median / mean_median } else { 0 } )
23f9fad4edfc planemo upload for repository https://github.com/HegemanLab/w4mcorcov_galaxy_wrapper/tree/master commit bd26542b811de06c1a877337a2840a9f899c2b94
eschen42
parents:
diff changeset
111 )
23f9fad4edfc planemo upload for repository https://github.com/HegemanLab/w4mcorcov_galaxy_wrapper/tree/master commit bd26542b811de06c1a877337a2840a9f899c2b94
eschen42
parents:
diff changeset
112 # "robust coefficient of variation, i.e., mad(feature-intensity for class-level max_level) / median(feature-intensity for class-level max_level)
23f9fad4edfc planemo upload for repository https://github.com/HegemanLab/w4mcorcov_galaxy_wrapper/tree/master commit bd26542b811de06c1a877337a2840a9f899c2b94
eschen42
parents:
diff changeset
113 salience_df$salient_rcv <- salience_df$max_rcv
23f9fad4edfc planemo upload for repository https://github.com/HegemanLab/w4mcorcov_galaxy_wrapper/tree/master commit bd26542b811de06c1a877337a2840a9f899c2b94
eschen42
parents:
diff changeset
114
23f9fad4edfc planemo upload for repository https://github.com/HegemanLab/w4mcorcov_galaxy_wrapper/tree/master commit bd26542b811de06c1a877337a2840a9f899c2b94
eschen42
parents:
diff changeset
115 return (salience_df)
23f9fad4edfc planemo upload for repository https://github.com/HegemanLab/w4mcorcov_galaxy_wrapper/tree/master commit bd26542b811de06c1a877337a2840a9f899c2b94
eschen42
parents:
diff changeset
116 }
23f9fad4edfc planemo upload for repository https://github.com/HegemanLab/w4mcorcov_galaxy_wrapper/tree/master commit bd26542b811de06c1a877337a2840a9f899c2b94
eschen42
parents:
diff changeset
117