CA_Makefile

BASENAME = p2
GENOME_SIZE = 4600000

SHELL = /bin/bash

CORRECTION_ROOT = /seq/schatz/a.thaliana/xiwang/james_workspace/nucmer-round-sim

TIGLEN_BIN = /sonas-hs/schatz/hpc/home/mschatz/build/bin/tig_length.pl 
PBTOOLS_BIN = /sonas-hs/schatz/hpc/home/gurtowsk/workspace/pbtools

JOIN_BIN = $(PBTOOLS_BIN)/join.py

CA_BIN = /sonas-hs/schatz/hpc/home/gurtowsk/sources/wgs.svn/Linux-amd64/bin
AMOS_BIN = /sonas-hs/schatz/hpc/home/gurtowsk/sources/amos/bin
MUMMER_BIN = /sonas-hs/schatz/hpc/home/gurtowsk/sources/MUMmer


STATS_BIN = $(AMOS_BIN)/stats

GATEKEP_BIN = $(CA_BIN)/gatekeeper
TIGSTOR_BIN = $(CA_BIN)/tigStore 
OVLSTOR_BIN = $(CA_BIN)/overlapStore

GTKSTOR = $(BASENAME).gkpStore
TIGSTOR = $(BASENAME).tigStore
OVLSTOR = $(BASENAME).ovlStore

.PHONY : all

all : unitigs.layout.clean.names unitigs.layout.stats

unitigs.layout :
	$(TIGSTOR_BIN) -g $(GTKSTOR) -t $(TIGSTOR) 1 -U -d layout > unitigs.layout

unitigs.layout.stats: unitigs.layout
	$(TIGLEN_BIN) unitigs.layout  | $(STATS_BIN) -f 2 -n50 $(GENOME_SIZE) > unitigs.layout.stats

ReadIndex.txt: 
	$(GATEKEP_BIN) -dumpfragments $(GTKSTOR) | grep "fragmentIdent" | \
	awk '{ split($$3,a,","); print a[2]"\t"a[1]}END{print "NONE\tNONE"}' > ReadIndex.txt

unitigs.layout.clean: unitigs.layout ReadIndex.txt
	cat unitigs.layout | \
	awk '{if($$1=="FRG"){$$1=$$2=$$3=$$4=""; print $$0}else if($$1=="unitig"){print}}' | \
	sed "s/^\s*//" | uniq  > unitigs.layout.clean

unitigs.layout.clean.names: unitigs.layout.clean
	python $(JOIN_BIN) ReadIndex.txt unitigs.layout.clean true > unitigs.layout.clean.names


utg_frg.db: unitigs.layout.clean
	cat unitigs.layout.clean  | \
	awk '{ if($$1=="unitig"){p = $$2}else{print $$1"\t"p}}' > utg_frg.db

frg.overlaps:
	$(OVLSTOR_BIN) -d5 -d3 -d $(OVLSTOR) > frg.overlaps

frg.utg.overlaps: frg.overlaps utg_frg.db
	paste <(python $(JOIN_BIN) utg_frg.db <(awk '{print $$1}' frg.overlaps )) \
	<(python $(JOIN_BIN) utg_frg.db <(awk '{print $$2}' frg.overlaps )) \
	<( awk '{print $$1"-"$$2}' frg.overlaps) > frg.utg.overlaps

frg.utg.overlaps.sorted: frg.utg.overlaps
	sort -n -k 1,1 -k 2,2 frg.utg.overlaps > frg.utg.overlaps.sorted


.PHONY : povl

povl :
	$(OVLSTOR_BIN) -p $(id) $(OVLSTOR) $(GTKSTOR) OBTCHIMERA


.PHONY : getsc

getsc:
	rn=$$(echo $(read) | awk '{split($$1,a,"_corrected"); print a[1]}') ; \
	idx=$$(grep "$$rn" $(CORRECTION_ROOT)/ReadIndex.txt | awk '{print $$2}'); \
	grep "$$rn" $(CORRECTION_ROOT)/$${idx}.delta.pre.r.sc


.PHONY : getrawsc

getrawsc:
	rn=$$(echo $(read) | awk '{split($$1,a,"_corrected"); print a[1]}') ; \
	idx=$$(grep "$$rn" $(CORRECTION_ROOT)/ReadIndex.txt | awk '{print $$2}'); \
	$(MUMMER_BIN)/show-coords -rclo $(CORRECTION_ROOT)/$${idx}.delta | grep "$$rn"

clean:
	rm -f unitigs.layout.clean.names unitigs.layout.clean \
	unitigs.layout.stats ReadIndex.txt unitigs.layout