From a29eb3e38414cd252850c9c4fb356f8b2bef6fb4 Mon Sep 17 00:00:00 2001
From: Christoph Helma <helma@in-silico.ch>
Date: Fri, 12 Feb 2021 19:54:07 +0100
Subject: model.rb refactored, mp2d models updated

---
 models/mutagenicity-mp2d/Makefile | 75 +++++++++++++++++++++++++++++++++++++++
 1 file changed, 75 insertions(+)
 create mode 100644 models/mutagenicity-mp2d/Makefile

(limited to 'models/mutagenicity-mp2d/Makefile')

diff --git a/models/mutagenicity-mp2d/Makefile b/models/mutagenicity-mp2d/Makefile
new file mode 100644
index 0000000..b80be6d
--- /dev/null
+++ b/models/mutagenicity-mp2d/Makefile
@@ -0,0 +1,75 @@
+SRC = Mutagenicity-Salmonella_typhimurium.csv 
+LAZAR_BIN = ../../bin
+MODEL = independent-variable-type dependent-variable-type dependent-variable-values similarity-thresholds independent-variables dependent-variables smiles metadata.json
+
+all: crossvalidation/summaries
+
+model: $(MODEL)
+
+# Crossvalidation
+
+crossvalidation/summaries: crossvalidation
+	$(LAZAR_BIN)/classification-summary.rb crossvalidation
+
+crossvalidation: $(MODEL)
+	$(LAZAR_BIN)/crossvalidation.rb .
+
+# Model
+
+independent-variables: smiles
+	$(LAZAR_BIN)/fingerprints.rb smiles > $@
+
+dependent-variables: $(SRC)
+	cut -f2 -d ',' $< | sed '1d' > $@
+
+smiles: $(SRC)
+	cut -f1 -d ',' $< | sed '1d' > $@
+
+# Metadata
+
+metadata.json:
+	echo '{"species":"Salmonella typhimurium","endpoint":"Mutagenicity","source":"http://cheminformatics.org/datasets/bursi/cas_4337.zip, http://doc.ml.tu-berlin.de/toxbenchmark/Mutagenicity_N6512.csv, https://data.europa.eu/euodp/data/storage/f/2017-07-19T142131/GENOTOX data and dictionary.xls","qmrf":{"group":"QMRF 4.10. Mutagenicity","name":"OECD 471 Bacterial Reverse Mutation Test"}}' > $@
+
+# Model parameters
+
+independent-variable-type:
+	echo "binary" > $@
+
+dependent-variable-type:
+	echo "binary" > $@
+
+dependent-variable-values:
+	echo -e "\"0\"\n\"1\"" > $@
+
+similarity-thresholds:
+	echo -e "0.2\n0.5" > $@
+
+# Merge data
+
+Mutagenicity-Salmonella_typhimurium.csv: download/efsa.csv download/hansen.csv download/kazius.csv
+	$(LAZAR_BIN)/merge_csv.rb $^
+
+# Convert data
+
+download/efsa.csv: download/efsa.tsv
+	download/efsa_classification.rb $< > $@
+
+download/efsa.tsv: download/efsa.xls
+	xls2csv -s cp1252 -d utf-8 -x -c "	" $< > $@
+
+download/hansen.csv: download/Mutagenicity_N6512.csv
+	download/hansen_convert.rb $< > $@
+
+download/kazius.csv: download/cas_4337.sdf
+	$(LAZAR_BIN)/sdf2csv.rb $< > $@
+
+# Download original data
+
+download/cas_4337.sdf:
+	cd download && wget http://cheminformatics.org/datasets/bursi/cas_4337.zip && unzip cas_4337.zip
+
+download/Mutagenicity_N6512.csv:
+	cd download && wget http://doc.ml.tu-berlin.de/toxbenchmark/Mutagenicity_N6512.csv
+
+download/efsa.xls:
+	wget "https://data.europa.eu/euodp/data/storage/f/2017-07-19T142131/GENOTOX data and dictionary.xls" -o $@
-- 
cgit v1.2.3