tensorflow cv with fixed mutagenicity outcome

1 files changed, 25 insertions, 24 deletions

diff --git a/Makefile b/Makefile
index 75e6e6d..e3279d3 100644
--- a/Makefile
+++ b/Makefile
@@ -15,12 +15,12 @@ TEMPLATE_FILE_LATEX   = pandoc-scholar.latex
 # crossvalidations
 
 LAZAR_CONFUSION_MATRIX_DIR = 10-fold-crossvalidations/lazar/crossvalidation/confusion_matrices
-LAZAR_PADEL_CONFUSION_MATRIX_DIR = 10-fold-crossvalidations/lazar-padel/crossvalidation/confusion_matrices
+LAZAR_PADEL_CONFUSION_MATRIX_DIR = 10-fold-crossvalidations/lazar-cdk/crossvalidation/confusion_matrices
 R_CV_DIR = 10-fold-crossvalidations/R
 TENSORFLOW_CV_DIR = 10-fold-crossvalidations/tensorflow
 
 CONFUSION_MATRICES_DIR = 10-fold-crossvalidations/confusion-matrices
-CONFUSION_MATRICES = $(CONFUSION_MATRICES_DIR)/lazar-all.csv $(CONFUSION_MATRICES_DIR)/lazar-high-confidence.csv $(CONFUSION_MATRICES_DIR)/lazar-padel-all.csv $(CONFUSION_MATRICES_DIR)/lazar-padel-high-confidence.csv $(CONFUSION_MATRICES_DIR)/R-RF.csv $(CONFUSION_MATRICES_DIR)/R-SVM.csv $(CONFUSION_MATRICES_DIR)/R-DL.csv $(CONFUSION_MATRICES_DIR)/tensorflow-rf.v3.csv $(CONFUSION_MATRICES_DIR)/tensorflow-lr.v3.csv $(CONFUSION_MATRICES_DIR)/tensorflow-lr2.v3.csv $(CONFUSION_MATRICES_DIR)/tensorflow-nn.v3.csv
+CONFUSION_MATRICES = $(CONFUSION_MATRICES_DIR)/lazar-all.csv $(CONFUSION_MATRICES_DIR)/lazar-high-confidence.csv $(CONFUSION_MATRICES_DIR)/lazar-cdk-all.csv $(CONFUSION_MATRICES_DIR)/lazar-cdk-high-confidence.csv $(CONFUSION_MATRICES_DIR)/R-RF.csv $(CONFUSION_MATRICES_DIR)/R-SVM.csv $(CONFUSION_MATRICES_DIR)/R-DL.csv $(CONFUSION_MATRICES_DIR)/tensorflow-rf-cdk.csv $(CONFUSION_MATRICES_DIR)/tensorflow-lr-cdk.csv $(CONFUSION_MATRICES_DIR)/tensorflow-lr2-cdk.csv $(CONFUSION_MATRICES_DIR)/tensorflow-nn-cdk.csv $(CONFUSION_MATRICES_DIR)/tensorflow-rf-mp2d.csv $(CONFUSION_MATRICES_DIR)/tensorflow-lr-mp2d.csv $(CONFUSION_MATRICES_DIR)/tensorflow-lr2-mp2d.csv $(CONFUSION_MATRICES_DIR)/tensorflow-nn-mp2d.csv
 
 CV_SUMMARY = 10-fold-crossvalidations/summary.yaml
 PA_SUMMARY = pyrrolizidine-alkaloids/summary.yaml
@@ -33,12 +33,12 @@ PA_LAZAR_DIR = $(PA_DIR)/lazar
 PA_R_DIR = $(PA_DIR)/R
 PA_TF_DIR = $(PA_DIR)/tensorflow
 
-PA_PREDICTIONS = $(PA_LAZAR_DIR)/pa-mp2d-predictions.csv $(PA_LAZAR_DIR)/pa-padel-predictions.csv $(PA_R_DIR)/PA.RF.outcome.csv $(PA_R_DIR)/PA.SVM.outcome.csv $(PA_R_DIR)/PA.DL.outcome.csv $(PA_TF_DIR)/pred.lr.v3-ext-Padel-2D.csv $(PA_TF_DIR)/pred.lr2.v3-ext-Padel-2D.csv $(PA_TF_DIR)/pred.rf.v3-ext-Padel-2D.csv $(PA_TF_DIR)/pred.nn.v3-ext-Padel-2D.csv
+PA_PREDICTIONS = $(PA_LAZAR_DIR)/pa-mp2d-predictions.csv $(PA_LAZAR_DIR)/pa-cdk-predictions.csv $(PA_R_DIR)/PA.RF.outcome.csv $(PA_R_DIR)/PA.SVM.outcome.csv $(PA_R_DIR)/PA.DL.outcome.csv $(PA_TF_DIR)/pred.lr.v3-ext-Padel-2D.csv $(PA_TF_DIR)/pred.lr2.v3-ext-Padel-2D.csv $(PA_TF_DIR)/pred.rf.v3-ext-Padel-2D.csv $(PA_TF_DIR)/pred.nn.v3-ext-Padel-2D.csv
 
 # manuscript
 
 TABLES = tables/lazar-summary.csv tables/r-summary.csv tables/tensorflow-summary.csv tables/pa-tab.tex tables/pa-summary.csv
-FIGURES = figures/roc.png figures/tsne-mp2d.png figures/tsne-padel.png
+FIGURES = figures/roc.png figures/tsne-mp2d.png figures/tsne-cdk.png
 
 all: $(TABLES) $(FIGURES) $(SUMMARY) mutagenicity.pdf 
 include $(PANDOC_SCHOLAR_PATH)/Makefile
@@ -48,11 +48,11 @@ mutagenicity.mustache.md: $(SUMMARY) mutagenicity.md $(TABLES) $(FIGURES)
 
 # figures
 
-figures/tsne-padel.png: figures/tsne-padel.csv
-	scripts/tsne-padel.R
+figures/tsne-cdk.png: figures/tsne-cdk.csv
+	scripts/tsne-cdk.R
 	
-figures/tsne-padel.csv: data/GenoTox-database.csv pyrrolizidine-alkaloids/PA-Padel-2D_m2.csv
-	scripts/padel-descriptors.rb $^ > $@
+figures/tsne-cdk.csv: data/GenoTox-database.csv pyrrolizidine-alkaloids/PA-Padel-2D_m2.csv
+	scripts/cdk-descriptors.rb $^ > $@
 	
 figures/tsne-mp2d.png: figures/tsne-mp2d.csv
 	scripts/tsne-mp2d.R
@@ -96,8 +96,8 @@ $(PA_SUMMARY): tables/pa-table.csv
 
 # PA predictions
 
-$(PA_LAZAR_DIR)/pa-padel-predictions.csv: 10-fold-crossvalidations/lazar-pa-padel/pa_independent_variables
-	../lazar/bin/batch_padel_classification.rb 10-fold-crossvalidations/lazar-pa-padel $< > $@
+$(PA_LAZAR_DIR)/pa-cdk-predictions.csv: 10-fold-crossvalidations/lazar-pa-cdk/pa_independent_variables
+	../lazar/bin/batch_padel_classification.rb 10-fold-crossvalidations/lazar-pa-cdk $< > $@
 
 $(PA_LAZAR_DIR)/pa-mp2d-predictions.csv: $(PA_LAZAR_DIR)/pa-smiles.csv
 	../lazar/bin/batch_fingerprint_classification.rb ../lazar/models/mutagenicity $< > $@
@@ -120,10 +120,10 @@ $(CONFUSION_MATRICES_DIR)/lazar-all.csv: $(LAZAR_CONFUSION_MATRIX_DIR)
 $(CONFUSION_MATRICES_DIR)/lazar-high-confidence.csv: $(LAZAR_CONFUSION_MATRIX_DIR)
 	cp $</high_confidence $@
 
-$(CONFUSION_MATRICES_DIR)/lazar-padel-all.csv: $(LAZAR_PADEL_CONFUSION_MATRIX_DIR)
+$(CONFUSION_MATRICES_DIR)/lazar-cdk-all.csv: $(LAZAR_PADEL_CONFUSION_MATRIX_DIR)
 	cp $</all $@
 
-$(CONFUSION_MATRICES_DIR)/lazar-padel-high-confidence.csv: $(LAZAR_PADEL_CONFUSION_MATRIX_DIR)
+$(CONFUSION_MATRICES_DIR)/lazar-cdk-high-confidence.csv: $(LAZAR_PADEL_CONFUSION_MATRIX_DIR)
 	cp $</high_confidence $@
 
 ## R
@@ -139,27 +139,28 @@ $(CONFUSION_MATRICES_DIR)/R-DL.csv: $(R_CV_DIR)/Sgl-Observations-DL.csv
 
 ## tensorflow
 
-$(TENSORFLOW_CV_DIR)/pred.lr.v3.norm.sorted.csv: $(TENSORFLOW_CV_DIR)/pred.lr.v3.norm.csv
-	sort -n $< > $@
+$(CONFUSION_MATRICES_DIR)/tensorflow-lr-cdk.csv: $(TENSORFLOW_CV_DIR)/pred.lr.v4.csv
+	scripts/cv-tensorflow-confusion-matrix.rb $< > $@
 
-$(TENSORFLOW_CV_DIR)/pred.lr2.v3.norm.sorted.csv: $(TENSORFLOW_CV_DIR)/pred.lr2.v3.norm.csv
-	sort -n $< > $@
+$(CONFUSION_MATRICES_DIR)/tensorflow-lr2-cdk.csv: $(TENSORFLOW_CV_DIR)/pred.lr2.v4.csv
+	scripts/cv-tensorflow-confusion-matrix.rb $< > $@
 
-$(TENSORFLOW_CV_DIR)/pred.nn.v3.norm.sorted.csv: $(TENSORFLOW_CV_DIR)/pred.nn.v3.norm.csv
-	sort -n $< > $@
+$(CONFUSION_MATRICES_DIR)/tensorflow-nn-cdk.csv: $(TENSORFLOW_CV_DIR)/pred.nn.v4.csv
+	scripts/cv-tensorflow-confusion-matrix.rb $< > $@
+
+$(CONFUSION_MATRICES_DIR)/tensorflow-rf-cdk.csv: $(TENSORFLOW_CV_DIR)/pred.rf.v4.csv
+	scripts/cv-tensorflow-confusion-matrix.rb $< > $@
 
-$(TENSORFLOW_CV_DIR)/pred.rf.v3.norm.sorted.csv: $(TENSORFLOW_CV_DIR)/pred.rf.v3.norm.csv
-	sort -n $< > $@
 
-$(CONFUSION_MATRICES_DIR)/tensorflow-lr.v3.csv: $(TENSORFLOW_CV_DIR)/pred.lr.v3.norm.sorted.csv
+$(CONFUSION_MATRICES_DIR)/tensorflow-lr-mp2d.csv: $(TENSORFLOW_CV_DIR)/pred.lr.v4_ext.csv
 	scripts/cv-tensorflow-confusion-matrix.rb $< > $@
 
-$(CONFUSION_MATRICES_DIR)/tensorflow-lr2.v3.csv: $(TENSORFLOW_CV_DIR)/pred.lr2.v3.norm.sorted.csv
+$(CONFUSION_MATRICES_DIR)/tensorflow-lr2-mp2d.csv: $(TENSORFLOW_CV_DIR)/pred.lr2.v4_ext.csv
 	scripts/cv-tensorflow-confusion-matrix.rb $< > $@
 
-$(CONFUSION_MATRICES_DIR)/tensorflow-nn.v3.csv: $(TENSORFLOW_CV_DIR)/pred.nn.v3.norm.sorted.csv
+$(CONFUSION_MATRICES_DIR)/tensorflow-nn-mp2d.csv: $(TENSORFLOW_CV_DIR)/pred.nn.v4_ext.csv
 	scripts/cv-tensorflow-confusion-matrix.rb $< > $@
 
-$(CONFUSION_MATRICES_DIR)/tensorflow-rf.v3.csv: $(TENSORFLOW_CV_DIR)/pred.rf.v3.norm.sorted.csv
+$(CONFUSION_MATRICES_DIR)/tensorflow-rf-mp2d.csv: $(TENSORFLOW_CV_DIR)/pred.rf.v4_ext.csv
 	scripts/cv-tensorflow-confusion-matrix.rb $< > $@