arks-make updated

arks-make.txt

#!/usr/bin/make -f
# Pipeline for the ARKS program
# Written by Jeffrey Tse

# w/ small updated by Brant Faircloth

#Default Parameters


# Input Names
draft=draft
reads=reads
# tigmint Parameters
minsize=2000
as=0.65
nm=5
dist=50000
mapq=0
trim=0
span=20
window=1000

# bwa Parameters
t=16
# ARKS Parameters
c=5
m=50-10000
z=500
j=0.55
k=30
r=0.05
e=30000
D=false
o=0
d=0
B=20
threads=16
# LINKS Parameters
l=5
a=0.3

SHELL=bash -e -o pipefail
ifeq ($(shell zsh -e -o pipefail -c 'true' 2>/dev/null; echo $$?), 0)
#Set pipefail to ensure that all commands of a pipe succeed.
SHELL=zsh -e -o pipefail
# Report run time and memory usage with zsh.
export REPORTTIME=1
export TIMEFMT=time user=%U system=%S elapsed=%E cpu=%P memory=%M job=%J
endif

# Record run time and memory usage in a file using GNU time.
ifdef time
ifneq ($(shell command -v gtime),)
gtime=command gtime -v -o [email protected]
else
gtime=command time -v -o [email protected]
endif
endif

.PHONY: all version help tigmint arks arks-tigmint
.DELETE_ON_ERROR:
.PRECIOUS: %_c$c_m$m_k$k_r$r_e$e_z$z.tigpair_checkpoint.tsv %_c$c_m$m_k$k_r$r_e$e_z$z_original.gv

all: help
# Help
help:
	@echo "Usage: ./arks-make [COMMAND] [OPTION=VALUE]..."
	@echo "    Commands:"
	@echo ""
	@echo "	arks            run arks only, skipping tigmint"
	@echo "	arks-tigmint    run tigmint, and run arks with output of tigmint"
	@echo "	help            display this help page"
	@echo "	version         display the software version"
	@echo " clean           remove intermediate files"
	@echo ""
	@echo "    General Options:"
	@echo ""
	@echo "	draft           draft name [draft]. File must have .fasta or .fa extension"
	@echo "	reads           read name [reads]. File must have .fastq.gz or .fq.gz extension"
	@echo "	time		logs time and memory usage to file for main steps (Set to 1 to enable logging)"
	@echo ""
	@echo "    Tigmint Options:"
	@echo ""
	@echo "	minsize         minimum molecule size [2000]"
	@echo "	as              minimum AS/read length ratio [0.65]"
	@echo "	nm              maximum number of mismatches [5]"
	@echo "	dist            max dist between reads to be considered the same molecule [50000]"
	@echo "	mapq            mapping quality threshold [0]"
	@echo "	trim            bp of contigs to trim after cutting at error [0]"
	@echo "	span            min number of spanning molecules to be considered assembled [20]"
	@echo "	window          window size for checking spanning molecules [1000]"
	@echo "	t               number of threads used [8]"
	@echo ""
	@echo "    ARKS Options:"
	@echo ""
	@echo "	c               minimum aligned read pairs per barcode mapping [5]"
	@echo "	m               barcode multiplicty range [50-10000]"
	@echo "	z               minimum contig length [500]"
	@echo "	k               k-mer size [30]"
	@echo "	j               min jaccard index for a read to be associated with a contigID [0.55]"
	@echo "	r               p-value for head/tail assigment and link orientation [0.05]"
	@echo "	e               contig head/tail length for masking aligments [30000]"
	@echo "	D               enable distance estimation [false]"
	@echo "	o               0 = no checkpoint files (default)"
	@echo "	                1 = produces output of kmerized draft"
	@echo "	                2 = produces output of aligning chromium to draft"
	@echo "	                3 = produces both 1 and 2"
	@echo "	B               estimate distance using N closest Jaccard scores [20]"
	@echo "	d               max node degree in scaffold graph [0]"
	@echo "	threads         numer of threads used for ARKS [1]"
	@echo ""
	@echo "    LINKS Options:"
	@echo ""
	@echo "	l               minimum number of links to compute scaffold [5]"
	@echo "	a               maximum link ratio between two best contig pairs [0.3]"
	@echo ""
	@echo "Example: To run tigmint and arks with myDraft.fa, myReads.fq.gz, and a custom multiplicity range, run:"
	@echo "	./arks-make arks-tigmint draft=myDraft reads=myReads m=[User defined multiplicty range]"
	@echo "To ensure that the pipeline runs correctly, make sure that the following tools are in your PATH: bwa, tigmint, samtools, arks (>= v1.0.2), LINKS (>= v1.8.6)"
# Version
version:
	@echo "arks-make 1.1"

clean:
	rm -f *.amb *.ann *.bwt *.pac *.sa *.dist.gv *.fai *.bed *.molecule.tsv *.sortbx.bam

#Preprocessing

# Create a .fa file that is soft linked to .fasta
%.fa: %.fasta
	ln -s $^ $@

# Create a .fq.gz file that is soft linked to .fastq.gz
%.fq.gz: %.fastq.gz
	ln -s $^ $@


#Run Tigmint
arks-tigmint: tigmint arks-with-tigmint

# Main
tigmint: $(draft).tigmint.fa


# Run tigmint
$(draft).tigmint.fa: $(draft).fa $(reads).fq.gz
	$(gtime) tigmint tigmint draft=$(draft) reads=$(reads) minsize=$(minsize) as=$(as) nm=$(nm) dist=$(dist) mapq=$(mapq) trim=$(trim) span=$(span) window=$(window) t=$t

#Run ARKS
arks: $(draft)_c$c_m$m_k$k_r$r_e$e_z$z_l$l_a$a.scaffolds.fa
arks-with-tigmint: $(draft).tigmint_c$c_m$m_k$k_r$r_e$e_z$z_l$l_a$a.scaffolds.fa

# Rename the draft headers
%.renamed.fa: %.fa
	perl -ne 'chomp; if(/>/){$$ct+=1; print ">$$ct\n";}else{print "$$_\n";} ' < $^ > $@

# Make fof file containing the reads filename
$(reads).fof: $(reads).fq.gz
	echo $^ > $@

# Create the barcode multiplicities file
$(reads)_multiplicities.csv: $(reads).fof
	calcBarcodeMultiplicities.pl $^ > $@

# Run ARKS Program
%_c$c_m$m_k$k_r$r_e$e_z$z_original.gv: %.renamed.fa $(reads)_multiplicities.csv $(reads).fq.gz
ifneq ($D, true)
	$(gtime) arks -p full -v -f $< -a $(word 2,$^) -c $c -t $(threads) -j $j -o $o -m $m -k $k -r $r -e $e -z $z -d $d -b $(patsubst %_original.gv,%,$@) $(word 3,$^)
else
	$(gtime) arks -p full -D -B $B -v -f $< -a $(word 2,$^) -c $c -t $(threads) -j $j -o $o -m $m -k $k -r $r -e $e -z $z -d $d -b $(patsubst %_original.gv,%,$@) $(word 3,$^)
endif

# Generate TSV from ARKS
%_c$c_m$m_k$k_r$r_e$e_z$z.tigpair_checkpoint.tsv: %_c$c_m$m_k$k_r$r_e$e_z$z_original.gv %.renamed.fa
	makeTSVfile.py $< $@ $(word 2,$^)

# Adds a and l paramters to the filename
%_c$c_m$m_k$k_r$r_e$e_z$z_l$l_a$a.tigpair_checkpoint.tsv: %_c$c_m$m_k$k_r$r_e$e_z$z.tigpair_checkpoint.tsv
	ln -s $^ $@

# Make an Empty fof File
empty.fof:
	touch $@

# Run LINKS
%_c$c_m$m_k$k_r$r_e$e_z$z_l$l_a$a.scaffolds.fa: %.renamed.fa empty.fof %_c$c_m$m_k$k_r$r_e$e_z$z_l$l_a$a.tigpair_checkpoint.tsv
	$(gtime) LINKS -f $< -s empty.fof -b $(patsubst %.scaffolds.fa,%,$@) -l $l -a $a -z $z