summaryrefslogtreecommitdiff
path: root/cv/scripts
diff options
context:
space:
mode:
Diffstat (limited to 'cv/scripts')
-rw-r--r--cv/scripts/CV_ds_pctype_algo_rseed.rb58
-rw-r--r--cv/scripts/factors_config17
-rwxr-xr-xcv/scripts/wrapper_pc_cv.sh45
3 files changed, 120 insertions, 0 deletions
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/cv/scripts/wrapper_pc_cv.sh b/cv/scripts/wrapper_pc_cv.sh
new file mode 100755
index 0000000..ed13122
--- /dev/null
+++ b/cv/scripts/wrapper_pc_cv.sh
@@ -0,0 +1,45 @@
+#!/bin/bash
+# Wrapper Skript for CV
+# Reads factors_config, ../data/datasets.yaml and performs cv's
+# Andreas Maunz, David Vorgrimmler, 2012
+
+if [ $# -lt 1 ]; then
+ echo "Usage: $0 factors"
+ exit
+fi
+
+PWD=`pwd`
+echo $PWD
+if [ ! -f $PWD/../data/datasets.yaml ]
+then
+ echo "datasets.yaml does not exist."
+ exit
+fi
+
+# Configure basics
+source $HOME/.bash_aliases
+otconfig
+THIS_DATE=`date +%Y%m%d_%H_`
+CV="CV_ds_pctype_algo_rseed.rb"
+FACTORS="$1"
+
+# Don't start when running
+while ps x | grep $CV | grep -v grep >/dev/null 2>&1; do sleep 3; done
+
+LOGFILE="$THIS_DATE""$USER""_wrapper_pc_cv.log"
+rm "$LOGFILE" >/dev/null 2>&1
+
+
+cat $FACTORS | while read factor; do
+ if ! [[ "$factor" =~ "#" ]]; then # allow comments
+ for r_seed in 1 #2 3 4 5
+ do
+ factor="$factor $r_seed"
+ echo "${THIS_DATE}: $factor" >> $LOGFILE>&1
+ echo "ruby $CV $factor" >> $LOGFILE 2>&1
+ ruby $CV $factor >> $LOGFILE 2>&1
+ echo >> $LOGFILE 2>&1
+ done
+ fi
+done
+