From b22ed5d6ce3812960023c40ea2bf0d836f159c0c Mon Sep 17 00:00:00 2001 From: David Vorgrimmler Date: Fri, 15 Jun 2012 18:47:49 +0200 Subject: Added min_frequency search. --- bbrc-sample/bbrc_sample_dv.rb | 77 +++++++++++++++++++++++++++++++++++ bbrc-sample/find_pm.rb | 95 ++++++++++++++++++++++++++++++++----------- 2 files changed, 148 insertions(+), 24 deletions(-) diff --git a/bbrc-sample/bbrc_sample_dv.rb b/bbrc-sample/bbrc_sample_dv.rb index 6a7c167..923469e 100644 --- a/bbrc-sample/bbrc_sample_dv.rb +++ b/bbrc-sample/bbrc_sample_dv.rb @@ -57,6 +57,83 @@ for i in 1..50 puts "[#{Time.now.iso8601(4).to_s}] Split result: #{datasets.to_yaml}" puts + # Find "good" min_frequency + + min_params = {} + min_params["dataset_uri"] = datasets[:training_ds] + + ds = OpenTox::Dataset.find(datasets[:training_ds]) +# ds_nr_de = ds.data_entries.size + ds_nr_com = ds.compounds.size + + min_params["backbone"] = backbone + durations = [] + x = ds_nr_com + ds_result_nr_f = 0 + y = x + y_old = 0 +# puts +# puts "----- Initialization: -----" + while ds_result_nr_f < (ds_nr_com/4).to_i do + y_old = y + y = x + x = (x/2).to_i + min_params["min_frequency"] = x +# puts "[#{Time.now.iso8601(4).to_s}] min_freq #{x}" + t = Time.now + result_uri = OpenTox::RestClientWrapper.post( File.join(CONFIG[:services]["opentox-algorithm"],"fminer/bbrc/"), min_params ) + durations << Time.now - t + ds_result = OpenTox::Dataset.find(result_uri) + ds_result_nr_f = ds_result.features.size +# puts "[#{Time.now.iso8601(4).to_s}] nr features #{ds_result_nr_f}" +# puts "[#{Time.now.iso8601(4).to_s}] duration #{durations.last}" +# puts "-------------" +# puts + end + +# puts "----- Main phase: -----" +# puts + max_duration = durations[0] +(ds_nr_com.to_f * 0.003) +# puts "max duration: #{max_duration}" +# puts + min_params["min_frequency"] = y + y = y_old + found = false + cnt = 0 + min_f = min_params["min_frequency"] + while found == false || cnt == 4 do + if min_f == min_params["min_frequency"] + cnt = cnt + 1 + end + min_f = min_params["min_frequency"] +# puts "[#{Time.now.iso8601(4).to_s}] min_freq #{min_params["min_frequency"]}" + t = Time.now + result_uri = OpenTox::RestClientWrapper.post( File.join(CONFIG[:services]["opentox-algorithm"],"fminer/bbrc/"), min_params ) + durations << Time.now - t + ds_result = OpenTox::Dataset.find(result_uri) + ds_result_nr_f = ds_result.features.size +# ds_result_nr_de = ds_result.data_entries.size +# ds_result_nr_com = ds_result.compounds.size +# puts "[#{Time.now.iso8601(4).to_s}] nr features #{ds_result_nr_f}" +# puts "[#{Time.now.iso8601(4).to_s}] duration #{durations.last}" +# puts "-------------" +# puts +# puts "smaller than #{(ds_nr_com*0.45).to_i} and larger than #{(ds_nr_com/10).to_i}" +# puts "x #{x}, y #{y}, min_freq #{min_params["min_frequency"]}" + if ds_result_nr_f.to_i < (ds_nr_com/2).to_i && ds_result_nr_f.to_i > (ds_nr_com/10).to_i + if durations.last < max_duration + found = true + min_freq = min_params["min_frequency"] + else + x = min_params["min_frequency"] + min_params["min_frequency"] = ((min_params["min_frequency"]+y)/2).to_i + end + else + y = min_params["min_frequency"] + min_params["min_frequency"] = ((x+min_params["min_frequency"])/2).to_i + end + end + # BBRC sample puts " ----- bbrc feature calulation -----" algo_params = {} diff --git a/bbrc-sample/find_pm.rb b/bbrc-sample/find_pm.rb index a541793..761e538 100644 --- a/bbrc-sample/find_pm.rb +++ b/bbrc-sample/find_pm.rb @@ -27,42 +27,89 @@ ds_name = ARGV[1] # e.g. MOU ds = YAML::load_file("#{path}") ds_uri = ds[ds_name]["dataset"] -algo_params = {} -algo_params["dataset_uri"] = ds_uri +min_params = {} +min_params["dataset_uri"] = ds_uri ds = OpenTox::Dataset.find(ds_uri) ds_nr_de = ds.data_entries.size ds_nr_com = ds.compounds.size + [true,false].each do |bb| - min_freq = 110 - duration = 0.0 - while duration < 60.0 && min_freq > 10 do - min_freq = min_freq - 10 - algo_params["min_frequency"] = min_freq.to_s + "pm" - algo_params["backbone"] = bb - #algo_params["nr_hits"] = false - #algo_params["feature_type"] = true - - - puts "[#{Time.now.iso8601(4).to_s}] #{algo_params.to_yaml}" + min_params["backbone"] = bb + #min_freq = 110 + durations = [] + x = ds_nr_com + ds_result_nr_f = 0 + y = x + y_old = 0 + puts + puts "----- Initialization: -----" + while ds_result_nr_f < (ds_nr_com/4).to_i do + y_old = y + y = x + x = (x/2).to_i + min_params["min_frequency"] = x + puts "[#{Time.now.iso8601(4).to_s}] min_freq #{x}" t = Time.now - result_uri = OpenTox::RestClientWrapper.post( File.join(CONFIG[:services]["opentox-algorithm"],"fminer/bbrc/"), algo_params ) - duration = Time.now - t + result_uri = OpenTox::RestClientWrapper.post( File.join(CONFIG[:services]["opentox-algorithm"],"fminer/bbrc/"), min_params ) + durations << Time.now - t ds_result = OpenTox::Dataset.find(result_uri) - ds_result_nr_de = ds_result.data_entries.size - ds_result_nr_com = ds_result.compounds.size ds_result_nr_f = ds_result.features.size - + puts "[#{Time.now.iso8601(4).to_s}] nr features #{ds_result_nr_f}" + puts "[#{Time.now.iso8601(4).to_s}] duration #{durations.last}" + puts "-------------" puts - puts "[#{Time.now.iso8601(4).to_s}] Bbrc result: #{result_uri}" - puts "[#{Time.now.iso8601(4).to_s}] nr dataentries: #{ds_result_nr_de} , (of #{ds_nr_de} ), #{(ds_result_nr_de/(ds_nr_de/100)).to_f.round}%" - puts "[#{Time.now.iso8601(4).to_s}] nr compounds: #{ds_result_nr_com} , (of #{ds_nr_com} ), #{(ds_result_nr_com/(ds_nr_com/100)).to_f.round}%" - puts "[#{Time.now.iso8601(4).to_s}] nr features: #{ds_result_nr_f}, , #{(ds_result_nr_f/(ds_nr_de/100)).to_f.round}%" - puts "[#{Time.now.iso8601(4).to_s}] Duration: #{duration}" - puts "------------------------" + end + puts "----- Main phase: -----" + puts + max_duration = durations[0] +(ds_nr_com.to_f * 0.003) + puts "max duration: #{max_duration}" + puts + min_params["min_frequency"] = y + y = y_old + found = false + cnt = 0 + min_f = min_params["min_frequency"] + while found == false || cnt == 4 do + if min_f == min_params["min_frequency"] + cnt = cnt + 1 + end + min_f = min_params["min_frequency"] + puts "[#{Time.now.iso8601(4).to_s}] min_freq #{min_params["min_frequency"]}" + t = Time.now + result_uri = OpenTox::RestClientWrapper.post( File.join(CONFIG[:services]["opentox-algorithm"],"fminer/bbrc/"), min_params ) + durations << Time.now - t + ds_result = OpenTox::Dataset.find(result_uri) + ds_result_nr_f = ds_result.features.size + ds_result_nr_de = ds_result.data_entries.size + ds_result_nr_com = ds_result.compounds.size + puts "[#{Time.now.iso8601(4).to_s}] nr features #{ds_result_nr_f}" + puts "[#{Time.now.iso8601(4).to_s}] duration #{durations.last}" + puts "-------------" puts + puts "smaller than #{(ds_nr_com*0.45).to_i} and larger than #{(ds_nr_com/10).to_i}" + puts "x #{x}, y #{y}, min_freq #{min_params["min_frequency"]}" + if ds_result_nr_f.to_i < (ds_nr_com/2).to_i && ds_result_nr_f.to_i > (ds_nr_com/10).to_i + if durations.last < max_duration + found = true + else + x = min_params["min_frequency"] + min_params["min_frequency"] = ((min_params["min_frequency"]+y)/2).to_i + end + else + y = min_params["min_frequency"] + min_params["min_frequency"] = ((x+min_params["min_frequency"])/2).to_i + end end + + puts + puts "[#{Time.now.iso8601(4).to_s}] Bbrc result: #{result_uri}" + puts "[#{Time.now.iso8601(4).to_s}] nr dataentries: #{ds_result_nr_de} , (of #{ds_nr_de} ), #{(ds_result_nr_de/(ds_nr_de/100)).to_f.round}%" + puts "[#{Time.now.iso8601(4).to_s}] nr compounds: #{ds_result_nr_com} , (of #{ds_nr_com} ), #{(ds_result_nr_com/(ds_nr_com/100)).to_f.round}%" + puts "[#{Time.now.iso8601(4).to_s}] nr features: #{ds_result_nr_f}, , #{(ds_result_nr_f/(ds_nr_de/100)).to_f.round}%" + puts "[#{Time.now.iso8601(4).to_s}] Duration: #{durations.last}" + puts "------------------------" puts end -- cgit v1.2.3