# Makefile for construction of databases that Jpred uses.
# The pdb make target has been over documented so as to provide
# explanation of the method behind the madness.
# All variables are defined at the top of the file, and then targets
# are defined afterwards

# Suffixes created by BLAST, ptm is also created, but I think this is
# a temporary file, so is mentioned explicitly in the clean target
BLAST_SUFFIX = .phr .pin .psd .psi .psq

# Specifically add path for some of the commands so we know we'll
# find them
PATH := $(PATH):/r0/www_servers/jpred/jpred_bin:/site/Linux/bin

# Define commands that occur multiple times
# The wget command means that you have to give the output location first,
# then the target
WGET = wget -N -t 10
FORMATDB = formatdb -o T -i

# Variables specific for PDB stuff
# URL to get the file from
PDB_URL = ftp://ftp.ebi.ac.uk/pub/databases/msd/pdb_aa.fasta
PDB_FILE = $(notdir $(PDB_URL))
# Target path to download to
PDB_FASTA = pdb.fasta
# Files created by running formatdb on $(PDB_FASTA)
PDB_BLAST = $(addprefix $(PDB_FASTA), $(BLAST_SUFFIX))
# Directory to install the files into
#PDB_TARGET = /r0/www_servers/jpred/databases/pdb
PDB_TARGET = /tmp/jpred/databases/pdb
# The target paths of the files that are to be installed
PDB_INSTALL =  $(addprefix $(PDB_TARGET)/, $(PDB_BLAST) $(PDB_FASTA))

# Variables specific for SWALL
SP_URL = ftp://ftp.ebi.ac.uk/pub/databases/sp_tr_nrdb/fasta/sprot.fas.gz
SP_FILE = $(notdir $(SP_URL))
TREMBL_URL = ftp://ftp.ebi.ac.uk/pub/databases/sp_tr_nrdb/fasta/trembl.fas.gz 
TREMBL_FILE = $(notdir $(TREMBL_URL))

SWALL_FASTA = swall.fasta
SWALL_INDX = swall.jidx
SWALL_FILT = swall.filt
SWALL_BLAST = $(addprefix $(SWALL_FILT), $(BLAST_SUFFIX))
SWALL_TARGET = /tmp/jpred/cluster/packages/jpred_swall
SWALL_INSTALL = $(addprefix $(SWALL_TARGET)/, $(SWALL_BLAST) $(SWALL_FASTA) $(SWALL_FILT))

######################################################################
all: pdb.db swall.db

######################################################################
# PDB FASTA seqeunce database

# Heirarchy of dependancies (indentation denotes dependancy):
# /installation/dir/pdb.fasta{,.{phr,pin,psd,psi,psq}}
# 	pdb_wget
# 	pdb.fasta{.{phr,pin,psd,psi,psq}}
#	 	pdb.fasta
# 			pdb_aa.fasta
# pdb_wget creates pdb_aa.fasta

# This states that the PDB target depends upon the pdb_wget target and
# the files in the variable $(PDB_INSTALL), this is the group of formatdb
# index files and the original FASTA file used to create them in the
# installation directory
pdb.db: pdb_wget $(PDB_INSTALL)

# This installs the files into the final directory
$(PDB_INSTALL): $(PDB_BLAST)
	mkdir -p $(PDB_TARGET)
	install -m 644 $(PDB_BLAST) $(PDB_FASTA) $(PDB_TARGET)

# Do BLAST DB formating. This depends on a group of files, so if any of
# them are deleted/too old, then they're reformated.
$(PDB_BLAST): $(PDB_FASTA)
	$(FORMATDB) $(PDB_FASTA)

$(PDB_FASTA): $(PDB_FILE)
	cp $(PDB_FILE) $(PDB_FASTA)

# This is a phony rule that's always executed, in this case it's to check
# whether a newer version of the PDB FASTA file has been deposited on the
# FTP site.
# It has to be a phony rule as if it's made a normal rule that depends
# on the file that's downloaded, its always considered upto date and the
# wget command isn't run.
.PHONY: pdb_wget
pdb_wget:
	$(WGET) $(PDB_FASTA) $(PDB_URL)
	
######################################################################

swall.db: sp_wget trembl_wget $(SWALL_INSTALL)

$(SWALL_INSTALL): $(SWALL_BLAST) $(SWALL_FASTA) $(SWALL_FILT) $(SWALL_INDX)
	mkdir -p $(SWALL_TARGET)
	install -m 644 $(SWALL_BLAST) $(SWALL_FASTA) $(SWALL_FILT) $(SWALL_INDX) $(SWALL_TARGET)

$(SWALL_BLAST): $(SWALL_FILT)
	$(FORMATDB) $(SWALL_FILT)

# This next line defines a temporary variable for the target
$(SWALL_FILT): TMP_FILE := $(shell tempfile -d /r0/scratch) 
$(SWALL_FILT): $(SWALL_FASTA)
	seg $(SWALL_FASTA) 12 2.2 2.5 -x > $(TMP_FILE)
	~/src/helixfilt/helixfilt $(TMP_FILE) > $(SWALL_FILT)
	rm $(TMP_FILE)

$(SWALL_INDX): $(SWALL_FASTA)
	db_dbmindex $(SWALL_FASTA) $(SWALL_INDX)

$(SWALL_FASTA): $(SP_FILE) $(TREMBL_FILE)
	gzip -c -d $(SP_FILE) > $(SWALL_FASTA)
	gzip -c -d $(TREMBL_FILE) >> $(SWALL_FASTA)

$(SP_FILE): sp_wget

$(TREMBL_FILE): trembl_wget

.PHONY: sp_wget
sp_wget:
	$(WGET) $(SP_URL)

.PHONY: trembl_wget
trembl_wget:
	$(WGET) $(TREMBL_URL)

######################################################################
clean:
	-rm $(SWALL_FASTA) $(SWALL_INDX) $(SWALL_FILT) $(SWALL_BLAST)
	-rm $(PDB_FASTA) $(PDB_BLAST)
	-rm formatdb.log *.ptm

distclean: clean
	-rm $(PDB_FILE) $(SP_FILE) $(TREMBL_FILE)

installclean:
	-rm -r $(SWALL_TARGET)
	-rm -r $(PDB_TARGET)