Makefile for downloading full-length amplicon sequencing MinION data from Benitez-Paez & Sanz: http://biorxiv.org/content/early/2017/06/08/117143

Makefile.amplicon_minion.benitez-paez_sanz

SHELL = /bin/bash

DDIR = orig_data
RDIR = results

PORETOOLS_BIN = . venv/bin/activate; poretools
PULLSEQ_BIN = ml vizbins_little_helpers; pullseq

MIN_LENGTH ?= 1000

#####
# BEAUTY TARGETS
#####
.PHONY: all

all:

setup_env: venv/bin/poretools
download_data: $(DDIR)/reads_R9.4.tar.gz
unzip_data: $(DDIR)/reads_R9.4/unzip.done
convert_to_fasta: $(RDIR)/read_R9.4/reads.fasta
size_select_fasta: $(RDIR)/read_R9.4/reads.ge_$(MIN_LENGTH)nt.fasta

#####
# ACTUAL TARGETS
#####
venv/bin/activate:
	@echo "##### SET-UP VIRTUALENV #####"
	@date
	virtualenv -p python2 venv
	@date
	@echo "##########"
	@echo

venv/bin/poretools: venv/bin/activate
	@echo "##### SET-UP PORETOOLS #####"
	@date
	. $^; pip install poretools
	@date
	@echo "##########"
	@echo

# $(DDIR)/reads.tar.gz:
# 	@echo "##### DOWNLOADING RAW DATA #####"
# 	@date
# 	mkdir -p $(dir $@)
# 	wget ftp://ftp.sra.ebi.ac.uk/vol1/ERA695/ERA695631/oxfordnanopore_native/reads.tar.gz -O $@
# 	@date
# 	@echo "##########"
# 	@echo

$(DDIR)/reads_R9.4.tar.gz:
	@echo "##### DOWNLOADING RAW DATA #####"
	@date
	mkdir -p $(dir $@)
	wget ftp://ftp.sra.ebi.ac.uk/vol1/ERA879/ERA879700/oxfordnanopore_native/reads_R9.4.tar.gz -O $@
	@date
	@echo "##########"
	@echo

$(DDIR)/reads_R9.4/unzip.done: $(DDIR)/reads_R9.4.tar.gz
	@echo "##### UNZIPPING RAW DATA #####"
	@date
	mkdir -p $(dir $@)
	tar -xzf $^ -C $(dir $@) && touch $@
	@date
	@echo "##########"
	@echo

$(RDIR)/read_R9.4/reads.fasta: $(DDIR)/reads_R9.4/unzip.done
	@echo "##### CONVERTING FAST5 TO FASTA #####"
	@date
	mkdir -p $(dir $@)
	$(PORETOOLS_BIN) fasta $(dir $^)/pass --type all > $@
	@date
	@echo "##########"
	@echo

%.ge_$(MIN_LENGTH)nt.fasta: %.fasta
	$(PULLSEQ_BIN) -i $^ -m $(MIN_LENGTH) > $@

#####
# CLEAN-UP
#####
clean:
	@echo "TODO: clean"

reads.tar.gz is commented out due to the large file size of 34 GB IIRC.
Might be interesting too look into those reads as well, yet they are based on (older) R9 chemistry, AFAIK.