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
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
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
|
require_relative 'task.rb'
require_relative 'prediction.rb'
require_relative 'helper.rb'
include OpenTox
configure :production do
$logger = Logger.new(STDOUT)
enable :reloader
end
configure :development do
$logger = Logger.new(STDOUT)
enable :reloader
end
before do
@version = File.read("VERSION").chomp
end
not_found do
redirect to('/predict')
end
error do
@error = request.env['sinatra.error']
haml :error
end
get '/?' do
redirect to('/predict')
end
get '/predict/?' do
if params[:tpid] && !params[:tpid].blank?
pid = params[:tpid].to_i
Process.kill(9,pid) #if (Process.getpgid(pid) rescue nil).present?
end
@models = Model::Validation.all
@models = @models.delete_if{|m| m.model.name =~ /\b(Net cell association)\b/}
@endpoints = @models.collect{|m| m.endpoint}.sort.uniq
@endpoints << "Oral toxicity (Cramer rules)"
@models.count <= 0 ? (haml :info) : (haml :predict)
end
get '/task/?' do
if params[:turi]
task = Task.find(params[:turi].to_s)
return JSON.pretty_generate(:percent => task.percent)
elsif params[:predictions]
pageSize = params[:pageSize].to_i - 1
pageNumber= params[:pageNumber].to_i - 1
compound = Compound.find @@compounds_ids[pageNumber]
image = compound.svg
smiles = compound.smiles
task = Task.find(params[:predictions].to_s)
unless task.predictions[params[:model]].nil?
if params[:model] == "Cramer"
prediction = task.predictions[params[:model]]
html = "<table class=\"table table-bordered single-batch\"><tr>"
html += "<td>#{image}</br>#{smiles}</br></td>"
string = "<td><table class=\"table\">"
string += "<tr class=\"hide-top\"><td>Cramer rules:</td><td>#{prediction["Cramer rules"][pageNumber.to_i]}</td>"
string += "<tr><td>Cramer rules, with extensions:</td><td>#{prediction["Cramer rules, with extensions"][pageNumber.to_i]}</td>"
string += "</table></td>"
html += "#{string}</tr></table>"
else
html = "<table class=\"table table-bordered single-batch\"><tr>"
html += "<td>#{image}</br>#{smiles}</br></td>"
string = "<td><table class=\"table\">"
prediction = task.predictions[params[:model]][pageNumber.to_i]
sorter = []
if prediction[:info]
sorter << {"Info" => prediction[:info]}
if prediction[:measurements_string].kind_of?(Array)
sorter << {"Measured activity" => "#{prediction[:measurements_string].join(";")}</br>#{prediction[:converted_measurements].join(";")}"}
else
sorter << {"Measured activity" => "#{prediction[:measurements_string]}</br>#{prediction[:converted_measurements]}"}
end
end
# regression
if prediction[:prediction_interval]
sorter << {"Prediction" => "#{prediction[:prediction_value]}</br>#{prediction[:converted_prediction_value]}"}
sorter << {"95% Prediction interval" => "#{prediction[:interval]}</br>#{prediction[:converted_interval]}"}
sorter << {"Warnings" => prediction[:warnings].join("</br>")}
# classification
elsif prediction[:probabilities]
sorter << {"Consensus prediction" => prediction["Consensus prediction"]}
sorter << {"Consensus confidence" => prediction["Consensus confidence"]}
sorter << {"Structural alerts for mutagenicity" => prediction["Structural alerts for mutagenicity"]}
sorter << {"Lazar mutagenicity (Salmonella typhimurium)" => ""}
sorter << {"Prediction" => prediction[:value]}
sorter << {"Probability" => prediction[:probabilities].collect{|k,v| "#{k}: #{v.signif(3)}"}.join("</br>")}
else
sorter << {"Warnings" => prediction[:warnings].join("</br>")}
end
sorter.each_with_index do |hash,idx|
k = hash.keys[0]
v = hash.values[0]
string += (idx == 0 ? "<tr class=\"hide-top\">" : "<tr>")+(k =~ /lazar/i ? "<td colspan=\"2\">" : "<td>")
# keyword
string += "#{k}:"
string += "</td><td>"
# values
string += "#{v}"
string += "</td></tr>"
end
string += "</table></td>"
html += "#{string}</tr></table>"
end
end
return JSON.pretty_generate(:predictions => [html])
end
end
get '/predict/modeldetails/:model' do
model = Model::Validation.find params[:model]
crossvalidations = Validation::RepeatedCrossValidation.find(model.repeated_crossvalidation_id).crossvalidations
return haml :model_details, :layout=> false, :locals => {:model => model, :crossvalidations => crossvalidations}
end
get '/jme_help/?' do
File.read(File.join('views','jme_help.html'))
end
get '/predict/dataset/:name' do
response['Content-Type'] = "text/csv"
dataset = Dataset.find_by(:name=>params[:name])
csv = dataset.to_csv
csv
end
get '/predict/csv/:task/:model/:filename/?' do
response['Content-Type'] = "text/csv"
task = Task.find params[:task].to_s
tempfile = Tempfile.new
tempfile.write(task.csv)
tempfile.rewind
send_file tempfile, :filename => "#{Time.now.strftime("%Y-%m-%d")}_lazar_batch_prediction_#{params[:model]}_#{params[:filename]}", :type => "text/csv", :disposition => "attachment"
end
post '/predict/?' do
# process batch prediction
if !params[:fileselect].blank?
if params[:fileselect][:filename] !~ /\.csv$/
bad_request_error "Please submit a csv file."
end
File.open('tmp/' + params[:fileselect][:filename], "w") do |f|
f.write(params[:fileselect][:tempfile].read)
end
@filename = params[:fileselect][:filename]
begin
input = Dataset.from_csv_file File.join("tmp", params[:fileselect][:filename]), true
$logger.debug "save dataset #{params[:fileselect][:filename]}"
if input.class == OpenTox::Dataset
@dataset = Dataset.find input
@compounds = @dataset.compounds
else
bad_request_error "Could not serialize file '#{@filename}'."
end
rescue
bad_request_error "Could not serialize file '#{@filename}'."
end
if @compounds.size == 0
message = dataset[:warnings]
@dataset.delete
bad_request_error message
end
@models = params[:selection].keys
# for single predictions in batch
@@compounds_ids = @compounds.collect{|c| c.id.to_s}
@tasks = []
@models.each{|m| t = Task.new; t.save; @tasks << t}
@predictions = {}
task = Task.run do
@models.each_with_index do |model,idx|
t = @tasks[idx]
unless model == "Cramer"
m = Model::Validation.find model
type = (m.regression? ? "Regression" : "Classification")
# add header for regression
if type == "Regression"
unit = (type == "Regression") ? "(#{m.unit})" : ""
converted_unit = (type == "Regression") ? "#{m.unit =~ /\b(mmol\/L)\b/ ? "(mg/L)" : "(mg/kg_bw/day)"}" : ""
header = "ID,Endpoint,Unique SMILES,inTrainingSet,Measurements,Prediction #{unit},Prediction #{converted_unit},"\
"Prediction Interval Low #{unit},Prediction Interval High #{unit},"\
"Prediction Interval Low #{converted_unit},Prediction Interval High #{converted_unit},"\
"inApplicabilityDomain,Note\n"
end
# add header for classification
if type == "Classification"
av = m.prediction_feature.accept_values
header = "ID,Endpoint,Unique SMILES,inTrainingSet,Measurements,Consensus Prediction,Consensus Confidence,"\
"Structural alerts for mutagenicity,Lazar Prediction,"\
"Lazar predProbability #{av[0]},Lazar predProbability #{av[1]},inApplicabilityDomain,Note\n"
end
# predict compounds
p = 100.0/@compounds.size
counter = 1
predictions = []
@compounds.each do |compound|
if Prediction.where(compound: compound.id, model: m.id).exists?
$logger.debug "prediction already exists !"
prediction = Prediction.find_by(compound: compound.id, model: m.id).prediction
else
prediction = m.predict(compound)
# save prediction object
prediction_object = Prediction.new
prediction_object[:compound] = compound.id
prediction_object[:model] = m.id
prediction_object[:prediction] = prediction
prediction_object.save
end
if type == "Classification"# consensus mutagenicity
sa_prediction = KaziusAlerts.predict(compound.smiles)
lazar_mutagenicity = prediction
confidence = 0
lazar_mutagenicity_val = (lazar_mutagenicity[:value] == "non-mutagenic" ? false : true)
if sa_prediction[:prediction] == false && lazar_mutagenicity_val == false
confidence = 0.85
elsif sa_prediction[:prediction] == true && lazar_mutagenicity_val == true
confidence = 0.85 * ( 1 - sa_prediction[:error_product] )
elsif sa_prediction[:prediction] == false && lazar_mutagenicity_val == true
confidence = 0.11
elsif sa_prediction[:prediction] == true && lazar_mutagenicity_val == false
confidence = ( 1 - sa_prediction[:error_product] ) - 0.57
end
prediction["Consensus prediction"] = sa_prediction[:prediction] == false ? "non-mutagenic" : "mutagenic"
prediction["Consensus confidence"] = confidence.signif(3)
prediction["Structural alerts for mutagenicity"] = sa_prediction[:matches].blank? ? "none" : sa_prediction[:matches].collect{|a| a.first}.join("; ")
end
# regression
unless prediction[:value].blank?
if type == "Regression"
prediction[:prediction_value] = "#{prediction[:value].delog10.signif(3)} #{unit}"
prediction["converted_prediction_value"] = "#{compound.mmol_to_mg(prediction[:value].delog10).signif(3)} #{converted_unit}"
end
end
unless prediction[:prediction_interval].blank?
interval = prediction[:prediction_interval]
prediction[:interval] = "#{interval[1].delog10.signif(3)} - #{interval[0].delog10.signif(3)} #{unit}"
prediction[:converted_interval] = "#{compound.mmol_to_mg(interval[1].delog10).signif(3)} - #{compound.mmol_to_mg(interval[0].delog10).signif(3)} #{converted_unit}"
end
prediction["unit"] = unit
prediction["converted_unit"] = converted_unit
if prediction[:measurements].is_a?(Array)
prediction["measurements_string"] = (type == "Regression") ? prediction[:measurements].collect{|value| "#{value.delog10.signif(3)} #{unit}"} : prediction[:measurements].join("</br>")
prediction["converted_measurements"] = prediction[:measurements].collect{|value| "#{compound.mmol_to_mg(value.delog10).signif(3)} #{unit =~ /mmol\/L/ ? "(mg/L)" : "(mg/kg_bw/day)"}"} if type == "Regression"
else
output["measurements_string"] = (type == "Regression") ? "#{prediction[:measurements].delog10.signif(3)} #{unit}}" : prediction[:measurements]
output["converted_measurements"] = "#{compound.mmol_to_mg(prediction[:measurements].delog10).signif(3)} #{(unit =~ /\b(mmol\/L)\b/) ? "(mg/L)" : "(mg/kg_bw/day)"}" if type == "Regression"
end
predictions << prediction.delete_if{|k,v| k =~ /neighbors|prediction_feature_id|r_squared|rmse/i}
t.update_percent((counter*p).ceil)
counter += 1
end
# write csv
t[:csv] = header + to_csv(model,predictions,@compounds)
# write predictions
@predictions["#{model}"] = predictions
else # Cramer model
#t[:csv] = to_csv(model,nil,@compounds)
compounds = @compounds.collect{|c| c.smiles}
prediction = [Toxtree.predict(compounds, "Cramer rules"), Toxtree.predict(compounds, "Cramer rules with extensions")]
output = {}
output["model_name"] = "Oral toxicity (Cramer rules)"
output["cramer_rules"] = prediction.collect{|array| array.collect{|hash| hash["Cramer rules"]}}.flatten.compact
output["cramer_rules_extensions"] = prediction.collect{|array| array.collect{|hash| hash["Cramer rules, with extensions"]}}.flatten.compact
# header
csv = "ID,Endpoint,Unique SMILES,Cramer rules,Cramer rules with extensions\n"
# content
compounds.each_with_index do |smiles, idx|
csv << "#{idx+1},#{output["model_name"]},#{smiles},"\
"#{output["cramer_rules"][idx] != "nil" ? output["cramer_rules"][idx] : "none" },"\
"#{output["cramer_rules_extensions"][idx] != "nil" ? output["cramer_rules_extensions"][idx] : "none"}\n"
end
predictions = {}
predictions["Cramer rules"] = output["cramer_rules"].collect{|rule| rule != "nil" ? rule : "none"}
predictions["Cramer rules, with extensions"] = output["cramer_rules_extensions"].collect{|rule| rule != "nil" ? rule : "none"}
# write csv
t[:csv] = csv
# write predictions
@predictions["#{model}"] = predictions
t.update_percent(100)
end
# save task
# append predictions as last action otherwise they won't save
# mongoid works with shallow copy via #dup
t[:predictions] = @predictions
t.save
end#models
end#main task
@pid = task.pid
$logger.debug "application pid: #{@pid}"
File.delete File.join("tmp", params[:fileselect][:filename])
return haml :batch
end
# single compound prediction
# validate identifier input
if !params[:identifier].blank?
@identifier = params[:identifier].strip
$logger.debug "input:#{@identifier}"
# get compound from SMILES
@compound = Compound.from_smiles @identifier
bad_request_error "'#{@identifier}' is not a valid SMILES string." if @compound.blank?
@models = []
@predictions = []
@toxtree = false
params[:selection].keys.each do |model_id|
if model_id == "Cramer"
@toxtree = true
@predictions << [Toxtree.predict(@compound.smiles, "Cramer rules"), Toxtree.predict(@compound.smiles, "Cramer rules with extensions")]
else
model = Model::Validation.find model_id
@models << model
if model.model.name =~ /kazius/
sa_prediction = KaziusAlerts.predict(@compound.smiles)
lazar_mutagenicity = model.predict(@compound)
confidence = 0
lazar_mutagenicity_val = (lazar_mutagenicity[:value] == "non-mutagenic" ? false : true)
if sa_prediction[:prediction] == false && lazar_mutagenicity_val == false
confidence = 0.85
elsif sa_prediction[:prediction] == true && lazar_mutagenicity_val == true
confidence = 0.85 * ( 1 - sa_prediction[:error_product] )
elsif sa_prediction[:prediction] == false && lazar_mutagenicity_val == true
confidence = 0.11
elsif sa_prediction[:prediction] == true && lazar_mutagenicity_val == false
confidence = ( 1 - sa_prediction[:error_product] ) - 0.57
end
@predictions << [lazar_mutagenicity, {:prediction => sa_prediction, :confidence => confidence}]
else
@predictions << model.predict(@compound)
end
end
end
haml :prediction
end
end
get '/style.css' do
headers 'Content-Type' => 'text/css; charset=utf-8'
scss :style
end
|