1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
|
require_relative "setup.rb"
class LazarPhyschemDescriptorTest < MiniTest::Test
def test_lazar_pc_descriptors
# check available descriptors
@descriptors = OpenTox::Algorithm::Descriptor::DESCRIPTORS.keys
assert_equal 111,@descriptors.size,"wrong number of physchem descriptors"
@descriptor_values = OpenTox::Algorithm::Descriptor::DESCRIPTOR_VALUES
# select descriptors for test
@num_features_offset = 0
@descriptors.keep_if{|x| x=~/^Openbabel\./}
@descriptors.delete("Openbabel.L5") # TODO Openbabel.L5 does not work, investigate!!!
unless defined?($short_tests)
# the actual descriptor calculation is rather fast, computing 3D structures takes time
# A CDK descriptor can calculate serveral values, e.g., ALOGP produces ALOGP.ALogP, ALOGP.ALogp2, ALOGP.AMR
# both is accepted (and tested here): Cdk.ALOGP (produces 3 features), or ALOGP.AMR (produces only 1 feature)
@descriptors += ["Cdk.ALOGP.AMR", "Cdk.WienerNumbers", "Joelib.LogP", "Joelib.count.HeteroCycles"]
@num_features_offset = 1 # Cdk.WienerNumbers produces 2 (instead of 1) features
end
puts "Descriptors: #{@descriptors}"
# UPLOAD DATA
@dataset = OpenTox::MeasuredDataset.from_csv_file File.join(DATA_DIR,"EPAFHM.medi.csv")
puts "Dataset: "+@dataset.id
@compound_smiles = "CC(C)(C)CN"
@compound_inchi = "InChI=1S/C5H13N/c1-5(2,3)4-6/h4,6H2,1-3H3"
prediction_a = build_model_and_predict(true)
prediction_b = build_model_and_predict(false)
assert_equal prediction_a,prediction_b,"predicted value differs depending on calculation method"
puts "Predicted value: #{prediction_a}"
# the actual value (from the complete EPAFHM dataset) is 5.45, but it is predicted higher when tested
# do not expect a fixed value, this might vary with, e.g., the calculated 3d structure by OB
assert prediction_a > 5,"predicted values should be above 5, is #{prediction_a}"
assert prediction_a < 15,"predicted values should be below 15, is #{prediction_a}"
end
def build_model_and_predict(precompute_feature_dataset=true)
model_params = {:dataset => @dataset}
#feat_gen_uri = File.join($algorithm[:uri],"descriptor","physchem")
if precompute_feature_dataset
# PRECOMPUTE FEATURES
p = "/tmp/mergedfile.csv"
f = File.open(p,"w")
f.puts File.read(File.join(DATA_DIR,"EPAFHM.medi.csv"))
f.puts "\"#{@compound_smiles}\","
f.close
d = OpenTox::Dataset.from_csv_file p
descriptors = OpenTox::Algorithm::Descriptor.physchem(d.compounds, @descriptors)
#model_params[:feature_dataset_uri] = OpenTox::Algorithm::Generic.new(feat_gen_uri).run({:dataset_uri => d.uri, :descriptors => @descriptors})
else
model_params[:feature_generation_uri] = feat_gen_uri
model_params[:descriptors] = @descriptors
end
# BUILD MODEL
#p descriptors
feature_dataset = OpenTox::CalculatedDataset.new
feature_dataset.compounds = @dataset.compounds
feature_dataset.data_entries = descriptors
feature_dataset.features = @descriptors.collect{|d| OpenTox::Feature.find_or_create_by(:title => d)}
feature_dataset["inchis"].each do |inchi|
assert_kind_of String, inchi
end
feature_dataset["feature_ids"].each do |id|
assert_kind_of BSON::ObjectId, id
end
feature_dataset.data_entries.each do |entry|
#p entry
assert_kind_of Array, entry
entry.each do |e|
#p e
assert_kind_of Float, e
end
end
feature_dataset.save
model = OpenTox::Model::Lazar.create @dataset, feature_dataset
#model = OpenTox::Model::Lazar.new model_uri
#assert_equal model_uri.uri?, true
#puts "Predicted variable: "+model.predicted_variable
# CHECK FEATURE DATASET
#feature_dataset_uri = model.metadata[RDF::OT.featureDataset].first
#puts "Feature dataset: #{feature_dataset_uri}"
#feature_dataset = OpenTox::Dataset.new(feature_dataset_uri)
assert_equal @dataset.compounds.size,feature_dataset.compounds.size,"Incorrect number of compounds in feature dataset"
features = feature_dataset.features
feature_titles = features.collect{|f| f.title}
@descriptors.each do |d|
if (d=~/^Cdk\./ and d.count(".")==1) # CDK descriptors (e.g. Cdk.ALOG are included as Cdk.ALOGP.ALogP, Cdk.ALOGP.ALogp2 ..)
match = false
feature_titles.each do |f|
match = true if f=~/d/
end
assert match,"feature not found #{d} in feature dataset #{feature_titles.inspect}"
else
assert feature_titles.include?(d),"feature not found #{d} in feature dataset #{feature_titles.inspect}"
end
end
assert_equal @descriptors.size,features.size,"Incorrect number of features in feature dataset"
#assert_equal (@descriptors.size+@num_features_offset),features.size,"wrong num features in feature dataset"
# predict compound
compound_uri = "#{$compound[:uri]}/#{@compound_inchi}"
compound = OpenTox::Compound.new @compound_inchi
prediction = model.predict :compound => compound
p prediction
#prediction = OpenTox::Dataset.new prediction_uri
#assert_equal prediction.uri.uri?, true
#puts "Prediction "+prediction.uri
# check prediction
assert prediction.features.collect{|f| f.uri}.include?(model.predicted_variable),"prediction feature #{model.predicted_variable} not included prediction dataset #{prediction.features.collect{|f| f.uri}}"
assert prediction.compounds.collect{|c| c.uri}.include?(compound_uri),"compound #{compound_uri} not included in prediction dataset #{prediction.compounds.collect{|c| c.uri}}"
assert_equal 1,prediction.compound_indices(compound_uri).size,"compound should only be once in the dataset"
prediction.data_entry_value(prediction.compound_indices(compound_uri).first,model.predicted_variable)
end
end
|
