From 7bbe4c444523f281d07f79aa8d0a4719668c3c80 Mon Sep 17 00:00:00 2001
From: Christoph Helma <helma@in-silico.ch>
Date: Sat, 20 Mar 2021 00:14:10 +0100
Subject: manuscript update

---
 Makefile | 66 +++++++++++++++++++++++++++++++++++++++++++++-------------------
 1 file changed, 47 insertions(+), 19 deletions(-)

(limited to 'Makefile')

diff --git a/Makefile b/Makefile
index a3ec0ee..8025943 100644
--- a/Makefile
+++ b/Makefile
@@ -7,7 +7,7 @@
 ARTICLE_FILE          = mutagenicity.mustache.md
 PANDOC_SCHOLAR_PATH   = pandoc-scholar
 OUTFILE_PREFIX        = mutagenicity
-DEFAULT_EXTENSIONS    = pdf
+DEFAULT_EXTENSIONS    = pdf docx
 PANDOC_WRITER_OPTIONS = --filter=pandoc-crossref --filter=pandoc-citeproc
 TEMPLATE_FILE_LATEX   = pandoc-scholar.latex
 
@@ -47,19 +47,26 @@ PA_SUMMARY = $(PA_DIR)summary.yaml
 
 # manuscript
 
-PA_FIGURES = $(addprefix figures/, \
-	Dehydropyrrolizidine.png \
-	Diester.png \
-	Macrocyclic.diester.png \
-	Monoester.png \
-	N.oxide.png \
-	Otonecine.png \
-	Platynecine.png \
-	Retronecine.png \
-	Tertiary.PA.png \
+TSNE_FIGURES = $(addprefix figures/tsne-, \
+	mp2d-mutagenicity.png \
+	cdk-mutagenicity.png \
+	mp2d-lazar-high-confidence-classifications.png \
+	mp2d-lazar-all-classifications.png \
+	mp2d-rf-classifications.png \
+	mp2d-lr-classifications.png \
+	mp2d-lr2-classifications.png \
+	mp2d-nn-classifications.png \
+	mp2d-svm-classifications.png \
+	cdk-lazar-high-confidence-classifications.png \
+	cdk-lazar-all-classifications.png \
+	cdk-rf-classifications.png \
+	cdk-lr-classifications.png \
+	cdk-lr2-classifications.png \
+	cdk-nn-classifications.png \
+	cdk-svm-classifications.png \
 )
 
-FIGURES = figures/roc.png figures/tsne-mp2d-mutagenicity.png figures/tsne-cdk-mutagenicity.png $(PA_FIGURES)
+FIGURES = figures/roc.png figures/pa-groups.png $(TSNE_FIGURES)
 
 DATA = data.yaml
 
@@ -77,6 +84,18 @@ $(DATA): $(PA_SUMMARY) $(CV_SUMMARY) mutagenicity/mutagenicity.csv mutagenicity/
 
 ## tsne
 
+figures/tsne-mp2d-%.png: figures/tsne-mp2d-coordinates.csv figures/mp2d-%
+	Rscript scripts/tsne-classifications.R $^ $@
+
+figures/mp2d-%-classifications: figures/tsne-mp2d-coordinates.csv crossvalidations/predictions/mp2d/%.csv pyrrolizidine-alkaloids/mp2d/%.csv
+	scripts/cv-pa-classifications.rb $^ > $@
+
+figures/tsne-cdk-%.png: figures/tsne-cdk-coordinates.csv figures/cdk-%
+	Rscript scripts/tsne-classifications.R $^ $@
+
+figures/cdk-%-classifications: figures/tsne-cdk-coordinates.csv crossvalidations/predictions/cdk/%.csv pyrrolizidine-alkaloids/cdk/%.csv
+	scripts/cv-pa-classifications.rb $^ > $@
+
 figures/tsne-%-mutagenicity.png: figures/tsne-%-coordinates.csv figures/tsne-%-mutagenicity.csv
 	Rscript scripts/tsne-mutagenicity.R $^ $@
 
@@ -104,15 +123,18 @@ figures/tsne-mp2d-distances.csv: mutagenicity/mutagenicity-mp2d pyrrolizidine-al
 ## roc
 
 figures/roc.png: figures/roc.csv
-	Rscript scripts/roc.R 
+	Rscript scripts/roc.R $< $@
 
 figures/roc.csv: $(CV_SUMMARY)
 	scripts/roc.rb $< > $@
 
 ## pa predictions per group
 
-$(PA_FIGURES): $(PA_DIR)pa-groups.csv $(PA_DIR)pa-predictions.csv
-	scripts/pa-groups.R $^
+figures/pa-groups.png: figures/pa-groups.csv
+	scripts/pa-groups.R $< $@
+
+figures/pa-groups.csv: $(PA_DIR)pa-groups.csv $(PA_DIR)pa-predictions.csv
+	scripts/pa-groups.rb $^ > $@
 
 # PA predictions
 
@@ -136,12 +158,15 @@ $(PA_DIR)pa-predictions.csv: $(PA_PREDICTIONS)
 
 ## predictions
 
-$(PA_DIR)%/lazar-all.csv: $(LAZAR_PREDICTIONS_DIR)pa-%/predictions 
+$(PA_DIR)%/lazar-all.csv: $(LAZAR_PREDICTIONS_DIR)pa-%/pa-%-prediction.csv
 	scripts/lazar-pa-predictions.rb $< > $@
 
-$(PA_DIR)%/lazar-high-confidence.csv: $(LAZAR_PREDICTIONS_DIR)pa-%/predictions 
+$(PA_DIR)mp2d/lazar-high-confidence.csv: $(LAZAR_PREDICTIONS_DIR)pa-mp2d/pa-mp2d-prediction.csv
 	scripts/lazar-pa-predictions.rb $< 0.5 > $@
 
+$(PA_DIR)cdk/lazar-high-confidence.csv: $(LAZAR_PREDICTIONS_DIR)pa-cdk/pa-cdk-prediction.csv
+	scripts/lazar-pa-predictions.rb $< 0.9 > $@
+
 $(PA_DIR)mp2d/%.csv: $(TENSORFLOW_PA_DIR)pred.%.v5-ext-ext-Padel-2D.csv $(PA_DIR)pa-cids.csv
 	scripts/tensorflow-pa-predictions.rb $^ > $@
 
@@ -167,12 +192,15 @@ $(CONFUSION_MATRICES_DIR)%: $(CV_PREDICTIONS_DIR)%
 
 ## predictions
 
-$(CV_PREDICTIONS_DIR)%/lazar-all.csv: $(LAZAR_MODELS_DIR)mutagenicity-%
+$(CV_PREDICTIONS_DIR)%/lazar-all.csv: $(LAZAR_MODELS_DIR)mutagenicity-%/crossvalidation/predictions.csv
 	scripts/lazar-cv-predictions.rb $< > $@
 
-$(CV_PREDICTIONS_DIR)%/lazar-high-confidence.csv: $(LAZAR_MODELS_DIR)mutagenicity-%
+$(CV_PREDICTIONS_DIR)mp2d/lazar-high-confidence.csv: $(LAZAR_MODELS_DIR)mutagenicity-mp2d/crossvalidation/predictions.csv
 	scripts/lazar-cv-predictions.rb $< 0.5 > $@
 
+$(CV_PREDICTIONS_DIR)cdk/lazar-high-confidence.csv: $(LAZAR_MODELS_DIR)mutagenicity-cdk/crossvalidation/predictions.csv
+	scripts/lazar-cv-predictions.rb $< 0.9 > $@
+
 $(CV_PREDICTIONS_DIR)mp2d/%.csv: $(TENSORFLOW_CV_DIR)pred.%.v4_ext.csv
 	scripts/tensorflow-cv-predictions.rb $< > $@
 
-- 
cgit v1.2.3