125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
|
# File 'lib/rbbt/expression_old/matrix.rb', line 125
def random_forest_importance(main, contrast = nil, field = nil, options = {})
features = Misc.process_options options, :features
features ||= []
path = Persist.persistence_path(matrix_file, {:dir => File.join(Matrix::MATRIX_DIR, 'random_forest_importance')}, {:main => main, :contrast => contrast, :field => field, :features => features})
Persist.persist(data, :tsv, :file => path, :no_load => false, :check => [matrix_file]) do
all_samples = labels.keys
main_samples = find_samples(main, field)
if contrast
contrast_samples = find_samples(contrast, field)
else
contrast_samples = all_samples - main_samples
end
main_samples = remove_missing(main_samples)
contrast_samples = remove_missing(contrast_samples)
TmpFile.with_file do |result|
R.run <<-EOF
library(randomForest);
orig = rbbt.tsv('#{matrix_file}');
main = c('#{main_samples * "', '"}')
contrast = c('#{contrast_samples * "', '"}')
features = c('#{features * "', '"}')
features = intersect(features, rownames(orig));
data = t(orig[features, c(main, contrast)])
data = cbind(data, Class = 0)
data[main, "Class"] = 1
rf = randomForest(factor(Class) ~ ., data, na.action = na.exclude)
rbbt.tsv.write(rf$importance, filename='#{ result }', key.field = '#{@key_field}')
EOF
TSV.open(result, :type => :single, :cast => :to_f)
end
end
end
|