Updated and moved scripts in new folder structure

5 files changed, 79 insertions, 102 deletions

diff --git a/5x_cv/CV_ds_pctype_prop_algo_rseed.rb b/5x_cv/CV_ds_pctype_prop_algo_rseed.rb
deleted file mode 100644
index b9c55dd..0000000
--- a/5x_cv/CV_ds_pctype_prop_algo_rseed.rb
+++ /dev/null
@@ -1,60 +0,0 @@
-# Do a 10-fold crossvalidation
-# # Author: Andreas Maunz, David Vorgrimmler
-# # @params: Dataset_name(see dataset_nestle.yaml), pc_type(electronic,cpsa or constitutional ...), prop(true or false), prediction_algorithm(local_mlr_prop or local_svm_regression ...)
-
-if ARGV.size != 5 
-  puts "Args: ds_name, pc_type, prop, algo, random_seed"
-  puts ARGV.size
-  exit
-end
-
-ds_file = "datasets_nestle.yaml"
-pwd=`pwd`
-path = "#{pwd.chop}/#{ds_file}"
-if File.exists?(path)
-  puts "#{ds_file} exists"
-else
-  puts "#{ds_file} does not exist."
-  exit
-end
-
-require 'rubygems'
-require 'opentox-ruby'
-require 'yaml'
-
-subjectid = nil
-
-ds_name = ARGV[0] # e.g. MOU
-pc_type = ARGV[1] # e.g. electronic,cpsa
-prop = ARGV[2]    # true or false
-algo = ARGV[3]    # e.g. local_svm_regression, local_mlr_prop
-r_seed = ARGV[4]  # 1, 2, ..., 10
-
-ds = YAML::load_file("datasets_nestle.yaml")
-ds_uri = ds[ds_name]["dataset"]
-pc_ds_uri = ds[ds_name][pc_type]
-
-algo_params = "pc_type=#{pc_type}";
-algo_params += ";feature_dataset_uri=#{pc_ds_uri}"
-algo_params += ";propositionalized=#{prop}"
-algo_params += ";prediction_algorithm=#{algo}"
-puts algo_params.to_yaml
-
-prediction_feature = OpenTox::Dataset.find(ds_uri).features.keys.first
-
-
-# Ready
-cv_args = {}
-cv_args[:dataset_uri] = ds_uri
-cv_args[:prediction_feature] = prediction_feature
-cv_args[:algorithm_uri] = "http://toxcreate3.in-silico.ch:8087/algorithm/lazar"
-cv_args[:algorithm_params] = algo_params
-cv_args[:stratified] = false
-cv_args[:random_seed] = r_seed
-puts cv_args.to_yaml
-
-cv = OpenTox::Crossvalidation.create(cv_args).uri
-puts cv
-
-cvr = OpenTox::CrossvalidationReport.create( cv , subjectid).uri
-puts cvr
diff --git a/5x_cv/factors_config b/5x_cv/factors_config
deleted file mode 100644
index 604ca45..0000000
--- a/5x_cv/factors_config
+++ /dev/null
@@ -1,38 +0,0 @@
-#Dataset pc_type prop prediction_algorithm
-MDD constitutional false local_svm_regression
-MDD constitutional true local_svm_regression
-MDD electronic,cpsa false local_svm_regression
-MDD electronic,cpsa true local_svm_regression
-MDD constitutional,electronic,cpsa false local_svm_regression
-MDD constitutional,electronic,cpsa true local_svm_regression
-#MDD constitutional true local_mlr_prop
-#MDD electronic,cpsa true local_mlr_prop
-#MDD constitutional,electronic,cpsa true local_mlr_prop
-FHM2 constitutional false local_svm_regression
-FHM2 constitutional true local_svm_regression
-FHM2 electronic,cpsa false local_svm_regression
-FHM2 electronic,cpsa true local_svm_regression
-FHM2 constitutional,electronic,cpsa false local_svm_regression
-FHM2 constitutional,electronic,cpsa true local_svm_regression
-#MOU constitutional true local_mlr_prop
-#MOU electronic,cpsa true local_mlr_prop
-#MOU constitutional,electronic,cpsa true local_mlr_prop
-RAT2 constitutional false local_svm_regression
-RAT2 constitutional true local_svm_regression
-RAT2 electronic,cpsa false local_svm_regression
-RAT2 electronic,cpsa true local_svm_regression
-RAT2 constitutional,electronic,cpsa false local_svm_regression
-RAT2 constitutional,electronic,cpsa true local_svm_regression
-#MOU constitutional true local_mlr_prop
-#MOU electronic,cpsa true local_mlr_prop
-#MOU constitutional,electronic,cpsa true local_mlr_prop
-MOU2 constitutional false local_svm_regression
-MOU2 constitutional true local_svm_regression
-MOU2 electronic,cpsa false local_svm_regression
-MOU2 electronic,cpsa true local_svm_regression
-MOU2 constitutional,electronic,cpsa false local_svm_regression
-MOU2 constitutional,electronic,cpsa true local_svm_regression
-#MOU constitutional true local_mlr_prop
-#MOU electronic,cpsa true local_mlr_prop
-#MOU constitutional,electronic,cpsa true local_mlr_prop
-
diff --git a/cv/scripts/CV_ds_pctype_algo_rseed.rb b/cv/scripts/CV_ds_pctype_algo_rseed.rb
new file mode 100644
index 0000000..36adb6f
--- /dev/null
+++ b/cv/scripts/CV_ds_pctype_algo_rseed.rb
@@ -0,0 +1,58 @@
+# Do a 10-fold crossvalidation
+# # Author: Andreas Maunz, David Vorgrimmler
+# # @params: Dataset_name(see dataset_nestle.yaml), pc_type(electronic,cpsa or constitutional ... or nil to disable), prediction_algorithm(local_mlr_prop or local_svm_regression ...)
+
+if ARGV.size != 4 
+  puts "Args: ds_name, pc_type, algo, random_seed"
+  puts ARGV.size
+  exit
+end
+
+ds_file = "datasets.yaml"
+pwd=`pwd`
+path = "#{pwd.chop}/../data/#{ds_file}"
+if File.exists?(path)
+  puts "#{ds_file} exists"
+else
+  puts "#{ds_file} does not exist."
+  exit
+end
+
+require 'rubygems'
+require 'opentox-ruby'
+require 'yaml'
+
+subjectid = nil
+
+ds_name = ARGV[0] # e.g. MOU
+pc_type = ARGV[1] # e.g. electronic,cpsa or nil to disable
+algo = ARGV[2]    # e.g. local_svm_regression, local_mlr_prop
+r_seed = ARGV[3]  # 1, 2, ..., 10
+
+ds = YAML::load_file("../data/datasets.yaml")
+ds_uri = ds[ds_name]["dataset"]
+pc_ds_uri = ds[ds_name][pc_type]
+
+algo_params = "prediction_algorithm=#{algo}"
+algo_params += ";pc_type=#{pc_type}" unless pc_type == "nil" 
+algo_params += ";feature_dataset_uri=#{pc_ds_uri}" unless pc_type == "nil" 
+puts algo_params.to_yaml
+
+prediction_feature = OpenTox::Dataset.find(ds_uri).features.keys.first
+
+
+# Ready
+cv_args = {}
+cv_args[:dataset_uri] = ds_uri
+cv_args[:prediction_feature] = prediction_feature
+cv_args[:algorithm_uri] = "http://toxcreate3.in-silico.ch:80XX/algorithm/lazar"
+cv_args[:algorithm_params] = algo_params
+cv_args[:stratified] = false
+cv_args[:random_seed] = r_seed
+puts cv_args.to_yaml
+
+#cv = OpenTox::Crossvalidation.create(cv_args).uri
+#puts cv
+
+#cvr = OpenTox::CrossvalidationReport.create( cv , subjectid).uri
+#puts cvr
diff --git a/cv/scripts/factors_config b/cv/scripts/factors_config
new file mode 100644
index 0000000..83bb54c
--- /dev/null
+++ b/cv/scripts/factors_config
@@ -0,0 +1,17 @@
+#Dataset pc_type prediction_algorithm
+MDD constitutional local_svm_regression
+MDD electronic,cpsa local_svm_regression
+MDD topological local_svm_regression
+MDD nil local_svm_regression
+#FHM constitutional local_svm_regression
+#FHM electronic,cpsa local_svm_regression
+#FHM topological local_svm_regression
+#FHM nil local_svm_regression
+#RAT constitutional local_svm_regression
+#RAT electronic,cpsa local_svm_regression
+#RAT topological local_svm_regression
+#RAT nil local_svm_regression
+#MOU constitutional local_svm_regression
+#MOU electronic,cpsa local_svm_regression
+#MOU topological local_svm_regression
+#MOU nil local_svm_regression
diff --git a/5x_cv/wrapper_pc_cv.sh b/cv/scripts/wrapper_pc_cv.sh
index 719885a..ed13122 100755
--- a/5x_cv/wrapper_pc_cv.sh
+++ b/cv/scripts/wrapper_pc_cv.sh
@@ -1,6 +1,6 @@
 #!/bin/bash
 # Wrapper Skript for CV
-# Reads factors_config, dataset_nestle.yaml and performs cv's
+# Reads factors_config, ../data/datasets.yaml and performs cv's
 # Andreas Maunz, David Vorgrimmler,  2012
 
 if [ $# -lt 1 ]; then
@@ -10,9 +10,9 @@ fi
 
 PWD=`pwd`
 echo $PWD
-if [ ! -f $PWD/datasets_nestle.yaml ] 
+if [ ! -f $PWD/../data/datasets.yaml ] 
 then
-  echo "datasets_nestle.yaml does not exist."
+  echo "datasets.yaml does not exist."
   exit
 fi
 
@@ -20,7 +20,7 @@ fi
 source $HOME/.bash_aliases
 otconfig
 THIS_DATE=`date +%Y%m%d_%H_`
-CV="CV_ds_pctype_prop_algo_rseed.rb"
+CV="CV_ds_pctype_algo_rseed.rb"
 FACTORS="$1"
 
 # Don't start when running