# Manuscript
# Requirements:
# pandoc-scholar (https://github.com/pandoc-scholar/pandoc-scholar) in PANDOC_SCHOLAR_PATH

# manuscript

ARTICLE_FILE          = mutagenicity.mustache.md
PANDOC_SCHOLAR_PATH   = pandoc-scholar
OUTFILE_PREFIX        = mutagenicity
DEFAULT_EXTENSIONS    = pdf docx
PANDOC_WRITER_OPTIONS = --filter=pandoc-crossref --filter=pandoc-citeproc
TEMPLATE_FILE_LATEX   = pandoc-scholar.latex

# experiments

# crossvalidations

CV_DIR = crossvalidations/
CV_PREDICTIONS_DIR = $(CV_DIR)predictions/

LAZAR_MODELS_DIR = ../lazar/models/
LAZAR_PREDICTIONS_DIR = ../lazar/predictions/

TENSORFLOW_CV_DIR = $(CV_DIR)/tensorflow/

CONFUSION_MATRICES_DIR = $(CV_DIR)confusion-matrices/

CV_FILES = lazar-all.csv lazar-high-confidence.csv rf.csv lr.csv lr2.csv nn.csv svm.csv

CONFUSION_MATRICES = $(addprefix $(CONFUSION_MATRICES_DIR)mp2d/, $(CV_FILES)) $(addprefix $(CONFUSION_MATRICES_DIR)cdk/, $(CV_FILES))

CV_PREDICTIONS = $(addprefix $(CV_PREDICTIONS_DIR)mp2d/, $(CV_FILES)) $(addprefix $(CV_PREDICTIONS_DIR)cdk/, $(CV_FILES))

CV_SUMMARY = $(CV_DIR)summary.yaml

# PA predictions

PA_DIR = pyrrolizidine-alkaloids/
TENSORFLOW_PA_DIR = $(PA_DIR)tensorflow/

PA_MP2D_LAZAR_DIR = $(PA_MP2D_DIR)lazar/
PA_CDK_LAZAR_DIR = $(PA_CDK_DIR)lazar/

PA_PREDICTIONS = $(addprefix $(PA_DIR)mp2d/, $(CV_FILES)) $(addprefix $(PA_DIR)cdk/, $(CV_FILES))

PA_SUMMARY = $(PA_DIR)summary.yaml

# manuscript

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/pa-groups.png $(TSNE_FIGURES)

DATA = data.yaml

all: mutagenicity.pdf $(PA_DIR)pa-predictions.pdf $(CV_PREDICTIONS) $(CONFUSION_MATRICES) $(PA_PREDICTIONS)
include $(PANDOC_SCHOLAR_PATH)/Makefile

mutagenicity.mustache.md: $(DATA) mutagenicity.md $(FIGURES)
	mustache $^ > $@

# manuscript data
$(DATA): $(PA_SUMMARY) $(CV_SUMMARY) mutagenicity/mutagenicity.csv mutagenicity/mutagenicity-cdk.csv $(PA_DIR)pa-predictions.csv
	scripts/data.rb $^ > $@

# figures

## 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 $^ $@

### factors

figures/tsne-%-mutagenicity.csv: figures/tsne-%-coordinates.csv mutagenicity/mutagenicity.csv
	scripts/tsne-mutagenicity.rb $^ > $@

### coordinates

figures/tsne-cdk-coordinates.csv: figures/tsne-cdk-descriptors.csv
	Rscript scripts/tsne-cdk-coordinates.R $< $@
	
figures/tsne-mp2d-coordinates.csv: figures/tsne-mp2d-distances.csv
	Rscript scripts/tsne-mp2d-coordinates.R $< $@

### input
	
figures/tsne-cdk-descriptors.csv:  mutagenicity/mutagenicity-cdk.csv pyrrolizidine-alkaloids/pa-cdk.csv
	scripts/tsne-cdk-descriptors.rb $^ > $@
	
figures/tsne-mp2d-distances.csv: mutagenicity/mutagenicity-mp2d pyrrolizidine-alkaloids/pa-mp2d
	scripts/tsne-mp2d-distances.rb $^ > $@

## roc

figures/roc.png: figures/roc.csv
	Rscript scripts/roc.R $< $@

figures/roc.csv: $(CV_SUMMARY)
	scripts/roc.rb $< > $@

## pa predictions per group

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

## summary

$(PA_SUMMARY): $(PA_DIR)pa-predictions.csv
	scripts/pa-summary.rb $< > $@

## pdf table

$(PA_DIR)pa-predictions.pdf: $(PA_DIR)pa-predictions.tex
	pdflatex -output-directory $(PA_DIR) $(PA_DIR)pa-predictions.tex

$(PA_DIR)pa-predictions.tex: $(PA_DIR)/pa-groups.csv $(PA_DIR)pa-predictions.csv
	scripts/pa-predictions-latex.rb $^ > $@

## table

$(PA_DIR)pa-predictions.csv: $(PA_PREDICTIONS)
	scripts/pa-predictions.rb $^ > $@

## predictions

$(PA_DIR)%/lazar-all.csv: $(LAZAR_PREDICTIONS_DIR)pa-%/pa-%-prediction.csv
	scripts/lazar-pa-predictions.rb $< > $@

$(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 $^ > $@

$(PA_DIR)cdk/%.csv: $(TENSORFLOW_PA_DIR)pred.%.v5-ext-Padel-2D.csv $(PA_DIR)pa-cids.csv
	scripts/tensorflow-pa-predictions.rb $^ > $@

## sanitize PA input data

$(PA_DIR)pa-ids.csv $(PA_DIR)pa-cids.csv $(PA_DIR)pa-names.tsv $(PA_DIR)pa-groups.csv $(PA_DIR)pa-cdk.csv: $(PA_DIR)src/180920_PA_complete_SMILES.csv $(PA_DIR)src/pa-groups.original.csv $(PA_DIR)src/PA-Padel-2D_m2.csv
	scripts/sanitize-pa-data.rb

# crossvalidation

## summary

$(CV_SUMMARY): $(CONFUSION_MATRICES)
	scripts/cv-summary.rb $^ > $@

## confusion matrices

$(CONFUSION_MATRICES_DIR)%: $(CV_PREDICTIONS_DIR)%
	scripts/confusion-matrix.rb $< > $@

## predictions

$(CV_PREDICTIONS_DIR)%/lazar-all.csv: $(LAZAR_MODELS_DIR)mutagenicity-%/crossvalidation/predictions.csv
	scripts/lazar-cv-predictions.rb $< > $@

$(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 $< > $@

$(CV_PREDICTIONS_DIR)cdk/%.csv: $(TENSORFLOW_CV_DIR)pred.%.norm.v4.csv
	scripts/tensorflow-cv-predictions.rb $< > $@

# cdk descriptors

mutagenicity/mutagenicity-cdk.csv: mutagenicity/src/mutagenicity-mod-2.new.csv
	cut -f1,3- -d ',' $< > $@

# mp2d fingerprints

mutagenicity/mutagenicity-mp2d: mutagenicity/mutagenicity.csv
	scripts/mp2d-fingerprints.rb $< > $@

$(PA_DIR)pa-mp2d: $(PA_DIR)pa-cids.csv
	scripts/mp2d-fingerprints.rb $<  > $@