diff --git a/.gitignore b/.gitignore
index 1a22bef91de95c3e3895a1481d166162daccc04a..87b9045e5ea387a3bbff850d38b2908e17066524 100644
--- a/.gitignore
+++ b/.gitignore
@@ -6,3 +6,4 @@
*.dot
*.png
*.html
+*.out
diff --git a/SConstruct.run b/SConstruct.run
index f72eaeb8869db30461986737e359f59ce2d73aad..166d31bf353f2c012a5b4ab3fcdd665b8543b747 100755
--- a/SConstruct.run
+++ b/SConstruct.run
@@ -1,4 +1,5 @@
#!/bin/bash
+set -xeu
LOGFILE="SConstruct.run.log."`date +%Y%m%d_%H%M%S`".out"
#echo "LOGFILE $LOGFILE"
python scons/scons.py -n --stack-size=1024 --diskcheck=none $@ 2>&1 | tee $LOGFILE
diff --git a/SConstruct.tree.sh b/SConstruct.tree.sh
index 8ec0be4571cf23a5ff81ef12a1c36f08fda5fc4c..baf0808232b9ea02cc455e2f0c26c94b93916919 100755
--- a/SConstruct.tree.sh
+++ b/SConstruct.tree.sh
@@ -4,7 +4,7 @@ rm SConstruct.tree.png 2>/dev/null
rm SConstruct.tree.gif 2>/dev/null
rm SConstruct.tree.html 2>/dev/null
-bash SConstruct.run -n --tree=prune,status $@ > SConstruct.tree.dat
+bash SConstruct.run -n --tree=prune,status $@ | grep -v '+-/' > SConstruct.tree.dat
#,derived
cat SConstruct.tree.dat | python SConstruct.graph.py $@
diff --git a/all.csv b/all.csv
new file mode 100644
index 0000000000000000000000000000000000000000..4c2e48e937010255b0c370e60b09aa9eddc96c8e
--- /dev/null
+++ b/all.csv
@@ -0,0 +1,13 @@
+#in fasta in csv
+sa1/sa.fa sa1/sa.csv
+#sa2/sa.fa sa2/sa.csv
+
+sh1/sh.fa sh1/sh.csv
+#sh2/sh.fa sh2/sh.csv
+
+#sl1/sl.fa sl1/sl.csv
+#sl2/sl.fa sl2/sl.csv
+
+sp1/sp.fa sp1/sp.csv
+#sp2/sp.fa sp2/sp.csv
+
diff --git a/run2.py b/run2.py
index 39e7fe47543063326730cea1bc45b89f94f39bab..9332f7bf9b529e2deacdd273eb0731d41916e6d8 100755
--- a/run2.py
+++ b/run2.py
@@ -25,7 +25,7 @@ MAP_FOLDER = 'map'
OUT_FOLDER = 'output'
GENOME_SIZE = 950000000
TMP_FOLDER = '/run/shm'
-TMP_PREFIX = os.path.basename(sys.argv[0])+'_'+str(os.getpid())+'_'
+TMP_PREFIX = 'bwa_scons_mapping_'+str(os.getpid())+'_'
FRC_EXE = '/home/assembly/nobackup/metrics/FRC_align/src/FRC'
BWA_EXE = '/usr/bin/bwa'
SAMTOOLS_EXE = 'samtools'
@@ -35,30 +35,20 @@ num_threads = 1
ONLY_PROCESS_IF_DONT_EXISTS = True
-CONTIG = None
-CSV_GROUP = None
-env = None
+CSV_PROJECT = None
+env = None
if __name__ == '__main__':
if len(sys.argv) < 3:
print "no contig given"
- print sys.argv[0], "<IN FASTA> <IN CSV CONFIG>"
+ print sys.argv[0], "<IN PROJECT CSV>"
sys.exit(1)
- CONTIG = sys.argv[1]
- CSV_GROUP = sys.argv[2]
+ CSV_PROJECT = sys.argv[1]
elif __name__ == 'SCons.Script':
- AddOption('--infasta',
- dest='CONTIG',
- type='string',
- nargs=1,
- action='store',
- metavar='INFASTA',
- help='input fasta'
- )
AddOption('--incsv',
- dest='CSV_GROUP',
+ dest='CSV_PROJECT',
type='string',
nargs=1,
action='store',
@@ -66,59 +56,39 @@ elif __name__ == 'SCons.Script':
help='input csv'
)
+ AddOption('--threads',
+ dest='num_threads',
+ type='int',
+ nargs=1,
+ action='store',
+ metavar='threads',
+ default=0,
+ help='number of threads'
+ )
+
os.environ['SHELL'] = '/bin/bash'
#env['SHELL'] = os.environ['SHELL']
- env = Environment( CONTIG=GetOption('CONTIG'), CSV_GROUP=GetOption('CSV_GROUP'), ENV = os.environ, SHELL='/bin/bash' )
-
- print "SHELL" , os.environ['SHELL']
- print "SCONS SHELL", env['SHELL']
+ env = Environment( CSV_PROJECT=GetOption('CSV_PROJECT'), ENV = os.environ, SHELL='/bin/bash' )
- CONTIG = env.GetOption('CONTIG')
- CSV_GROUP = env.GetOption('CSV_GROUP')
+ CSV_PROJECT = env.GetOption('CSV_PROJECT')
+ if CSV_PROJECT is None:
+ print "input project csv not defined"
+ print "scons --incsv=<IN PROJECT FILE>"
+ sys.exit(1)
+ print "SHELL" , os.environ['SHELL']
+ print "SCONS SHELL", env['SHELL']
-if CONTIG is None:
- print "no fasta file given: %s --infasta <fasta> --incsv <csv> [-j <num threads>]" % sys.argv[0]
- sys.exit(1)
-
-if CSV_GROUP is None:
- print "no csv file given: %s --infasta <fasta> --incsv <csv> [-j <num threads>]" % sys.argv[0]
- sys.exit(1)
-
-if (not os.path.exists( CONTIG )) or (not os.path.exists( CSV_GROUP )):
- if (not os.path.exists( CONTIG )):
- print "contig file %s does not exists" % CONTIG
- if (not os.path.exists( CSV_GROUP )):
- print "csv group file %s does not exists" % CSV_GROUP
- print sys.argv[0], "<IN FASTA> <IN CSV GROUP>"
+if not os.path.exists(CSV_PROJECT):
+ print "input project csv %s does not exists" % CSV_PROJECT
sys.exit(1)
-outfolder = os.path.dirname( os.path.abspath(CSV_GROUP) )
-DB_FOLDER = os.path.join(outfolder, DB_FOLDER )
-MAP_FOLDER = os.path.join(outfolder, MAP_FOLDER )
-OUT_FOLDER = os.path.join(outfolder, OUT_FOLDER )
-
-print "BASE FOLDER: %s" % outfolder
-print "DB FOLDER: %s" % DB_FOLDER
-print "MAP FOLDER: %s" % MAP_FOLDER
-print "OUT FOLDER: %s" % OUT_FOLDER
-
-
-#RND_PREFIX = os.path.join(TMP_FOLDER, os.path.basename(CONTIG) + os.path.basename(CSV_GROUP))
-if not dry_run:
- if not os.path.exists(DB_FOLDER):
- os.makedirs(DB_FOLDER)
- if not os.path.exists(MAP_FOLDER):
- os.makedirs(MAP_FOLDER)
-
- if not os.path.exists(OUT_FOLDER):
- os.makedirs(OUT_FOLDER)
def my_decider(dependency, target, prev_ni):
@@ -151,19 +121,23 @@ if __name__ == 'SCons.Script':
# How many CPU's can we use ?
#env.SetOption("SHELL", "/bin/bash")
- num_cpu = multiprocessing.cpu_count()
- num_jobs = int( env.GetOption('num_jobs') )
- print "num cpu %d" % num_cpu
- print "num jobs %d" % num_jobs
+ num_cpu = multiprocessing.cpu_count()
+ num_jobs = int( env.GetOption('num_jobs' ) )
+ num_threads = int( env.GetOption('num_threads') )
+ print "num cpu %d" % num_cpu
+ print "num jobs %d" % num_jobs
+ print "num threads %d" % num_threads
if num_jobs <= 1:
- num_threads = num_cpu/4
- env.SetOption('num_jobs', num_threads)
+ num_jobs = int((num_cpu/8) + 1)
- else:
- num_threads = num_jobs
+ if num_threads == 0:
+ num_threads = (num_jobs*2)
+
+ env.SetOption('num_jobs' , num_jobs )
- print "running with -j", env.GetOption('num_jobs')
+ print "running with %d jobs" % num_jobs
+ print "running with %d threads" % num_threads
# Set number of simultaneous threads to something less than the number of CPUs
# /CPUs
@@ -257,7 +231,7 @@ def move(*args, **kwargs):
##########################
## ADD COMMAND WRAPPER
##########################
-#addCommand('bwa indexing', CMD_BWA_INDEX, TARGET = bwt_index, SOURCE = CONTIG, DEPS = [] })
+#addCommand('bwa indexing', CMD_BWA_INDEX, TARGET = bwt_index, SOURCE = setup['in_fasta'], DEPS = [] })
def addCommand(name, cmd, TARGET=None, SOURCE=[], DATA={}, DEPS=[], REQS=[], PRECIOUS=False, USETMP=False, TMP_SUFFIX=""):
if TARGET is None:
print "no target defined:", str(locals())
@@ -334,17 +308,20 @@ def addCommand(name, cmd, TARGET=None, SOURCE=[], DATA={}, DEPS=[], REQS=[], PRE
- pdata = { 'SOURCE': SOURCE, 'TARGET': TARGET, 'DEPS': DEPS, 'REQS': REQS }
+ pdata = { 'SOURCE': SOURCE, 'TARGET': TARGET, 'RESULTS': TARGET, 'DEPS': DEPS, 'REQS': REQS }
+
datan = copy.copy(DATA)
for key in pdata:
datan[key] = pdata[key]
+
+
+
outFile = None
tmpFile = None
- if USETMP:
+ if (not allOk) and USETMP:
tmpFile = tempfile.mktemp(dir=TMP_FOLDER, prefix=TMP_PREFIX, suffix=TMP_SUFFIX)
- print " USING TMP :: %s as %s" % ( datan['TARGET'], tmpFile )
- outFile = datan['TARGET']
+ print " USING TMP :: %s as %s" % ( datan['TARGET'][0], tmpFile )
datan['TARGET'] = tmpFile
@@ -372,8 +349,6 @@ def addCommand(name, cmd, TARGET=None, SOURCE=[], DATA={}, DEPS=[], REQS=[], PRE
elif __name__ == 'SCons.Script':
print " adding"
- datan['RESULTS'] = datan['TARGET']
-
if ONLY_PROCESS_IF_DONT_EXISTS:
if allOk:
print " all ok. skipping"
@@ -386,26 +361,22 @@ def addCommand(name, cmd, TARGET=None, SOURCE=[], DATA={}, DEPS=[], REQS=[], PRE
else:
print " not ok. adding"
- datan['TARGET'] = env.Command(datan['TARGET'], datan['SOURCE'], cmdCall)
+ env.Command(datan['TARGET'], datan['SOURCE'], cmdCall)
else:
print " appending always"
- datan['RESULTS'] = env.Command(datan['TARGET'], datan['SOURCE'], cmdCall)
+ env.Command(datan['TARGET'], datan['SOURCE'], cmdCall)
- for dp in DEPS: env.Depends( datan['RESULTS'], dp )
- for rq in REQS: env.Requires( datan['RESULTS'], rq )
+ if (not allOk) and USETMP:
+ print " USING TMP: %s to %s" % (datan['TARGET'], datan['RESULTS'][0])
+ env.Command( datan['RESULTS'][0] , datan['TARGET'], move )
+ for dp in DEPS: env.Depends( datan['TARGET'], dp )
+ for rq in REQS: env.Requires( datan['TARGET'], rq )
- if USETMP:
- print " USING TMP: %s to %s" % (datan['RESULTS'][0], outFile[0] )
- env.Command( outFile , datan['RESULTS'], move )
- #env.Depends( datan['RESULTS'], outFile )
- if PRECIOUS: env.Precious( outFile )
-
- else:
- if PRECIOUS: env.Precious( datan['RESULTS'] )
+ if PRECIOUS: env.Precious( datan['RESULTS'] )
print "\n\n"
@@ -440,14 +411,91 @@ def run_cmd( CMD , name ):
print " DRY RUN:: skipping %s :: %s\n" % ( name , CMD )
return 0
-
def main():
+ setup = []
+ with open(CSV_PROJECT, 'r') as csvfhd:
+ for line in csvfhd:
+ line = line.strip()
+
+ if len(line) == 0: continue
+ if line[0] == "#": continue
+
+ in_fasta, in_csv = line.split("\t")
+
+ if not os.path.exists(in_fasta):
+ print "input fasta %s does not exists" % in_fasta
+ sys.exit(1)
+
+ if not os.path.exists(in_csv):
+ print "input csv %s does not exists" % in_csv
+ sys.exit(1)
+
+
+ base_folder = os.path.dirname( os.path.abspath(in_csv) )
+ proj_name = os.path.basename(base_folder)
+ db_folder = os.path.join(base_folder, DB_FOLDER )
+ map_folder = os.path.join(base_folder, MAP_FOLDER )
+ out_folder = os.path.join(base_folder, OUT_FOLDER )
+
+ if 'all' not in COMMAND_LINE_TARGETS:
+ if proj_name not in COMMAND_LINE_TARGETS:
+ print "project %s not to be built. skipping parsing" % proj_name
+ continue
+
+ print "PROJECT NAME: %s" % proj_name
+ print "FASTA FILE : %s" % in_fasta
+ print "CSV FILE : %s" % in_csv
+ print "BASE FOLDER: %s" % base_folder
+ print "DB FOLDER: %s" % db_folder
+ print "MAP FOLDER: %s" % map_folder
+ print "OUT FOLDER: %s" % out_folder
+ print
+
+ if not dry_run:
+ if not os.path.exists(db_folder):
+ os.makedirs(db_folder)
+
+ if not os.path.exists(map_folder):
+ os.makedirs(map_folder)
+
+ if not os.path.exists(out_folder):
+ os.makedirs(out_folder)
+
+ projsetup = {
+ 'proj_name' : proj_name,
+ 'in_fasta' : in_fasta,
+ 'in_csv' : in_csv,
+ 'base_folder': base_folder,
+ 'db_folder' : db_folder,
+ 'map_folder' : map_folder,
+ 'out_folder' : out_folder
+ }
+ pp(projsetup)
+ setup.append(projsetup)
+
+ projout = []
+ for projsetup in setup:
+ projout.extend( project(projsetup) )
+
+ projout = [ x for x in sorted(list(set(projout))) if x is not None ]
+
+ if __name__ == 'SCons.Script':
+
+ env.Alias( 'all', projout )
+
+ print "target ALL:", "\n ".join( projout )
+ for projsetup in setup:
+ print "target %s: %s" % (projsetup['proj_name'].upper(), projsetup['outfile'])
+ print "\n\n\n"
+
+def project( setup ):
+
datas = {}
##########################
## READ CSV
##########################
- with open(CSV_GROUP, 'r') as fhd:
+ with open(setup['in_csv'], 'r') as fhd:
line_count = 0
for line in fhd:
line_count += 1
@@ -513,9 +561,9 @@ def main():
for plat in sorted(datas):
data = datas[plat]
if plat == 'illumina':
- datas[plat] = processIllumina(data)
+ datas[plat] = processIllumina(data, setup)
elif plat == '454' :
- datas[plat] = process454(data)
+ datas[plat] = process454(data, setup)
@@ -538,10 +586,11 @@ def main():
dataf[name] = datas[plat][name]
- run_bwa(dataf)
+ finalout = run_bwa(dataf, setup)
+ return finalout
-def process454(data):
+def process454(data, setup):
for comm_name in sorted(data):
print "parsing 454 :: %s" % comm_name
@@ -554,7 +603,7 @@ def process454(data):
sys.exit(1)
- out_prefix = os.path.join( MAP_FOLDER , comm_name )
+ out_prefix = os.path.join( setup['map_folder'] , comm_name )
print "parsing 454 :: %s :: %s :: out prefix %s" % ( comm_name, infile, out_prefix )
out_fastq = out_prefix + '.fastq'
@@ -581,11 +630,11 @@ def process454(data):
addCommand('extracting sff', CMD_EXTRACT, TARGET = data[comm_name]['file_names_454_q'], SOURCE = data[comm_name]['file_names_454' ], DATA = data[comm_name], REQS = [ ])
addCommand('splitting sff' , CMD_SPLIT , TARGET = data[comm_name]['file_names' ], SOURCE = data[comm_name]['file_names_454_q'], DATA = data[comm_name], REQS = [ ])
- data = processIllumina(data)
+ data = processIllumina(data, setup)
return data
-def processIllumina(data):
+def processIllumina(data, setup):
##########################
## GENERATE FILE NAMES
##########################
@@ -594,18 +643,18 @@ def processIllumina(data):
RND_NAME2 = tempfile.mktemp(dir=TMP_FOLDER, prefix='bwa_')
- out_sam = os.path.join( MAP_FOLDER , comm_name + '.sam' )
- out_bam_tmp = os.path.join( MAP_FOLDER , comm_name + '.bam' )
+ out_sam = os.path.join( setup['map_folder'] , comm_name + '.sam' )
+ out_bam_tmp = os.path.join( setup['map_folder'] , comm_name + '.bam' )
- out_bam_pfx = os.path.join( OUT_FOLDER , comm_name )
- out_bam = os.path.join( OUT_FOLDER , comm_name + '.bam' )
+ out_bam_pfx = os.path.join( setup['out_folder'] , comm_name )
+ out_bam = os.path.join( setup['out_folder'] , comm_name + '.bam' )
out_bai = out_bam + '.bai'
- out_cov = os.path.join( OUT_FOLDER , comm_name + '.cov' )
- out_stats = os.path.join( OUT_FOLDER , comm_name + '.stats' )
- out_flag = os.path.join( OUT_FOLDER , comm_name + '.flagstats')
- out_ok = os.path.join( OUT_FOLDER , comm_name + '.ok' )
- out_frc = os.path.join( OUT_FOLDER , comm_name + '_FRC__CEstats_PE.txt')
- out_frcB = os.path.join( OUT_FOLDER , comm_name + '_FRC_' )
+ out_cov = os.path.join( setup['out_folder'] , comm_name + '.cov' )
+ out_stats = os.path.join( setup['out_folder'] , comm_name + '.stats' )
+ out_flag = os.path.join( setup['out_folder'] , comm_name + '.flagstats')
+ out_ok = os.path.join( setup['out_folder'] , comm_name + '.ok' )
+ out_frc = os.path.join( setup['out_folder'] , comm_name + '_FRC__CEstats_PE.txt')
+ out_frcB = os.path.join( setup['out_folder'] , comm_name + '_FRC_' )
filenames = data[comm_name]['file_names']
@@ -634,11 +683,11 @@ def processIllumina(data):
data[comm_name]['file_names_str' ] = " ".join( filenames )
data[comm_name]['fastq1' ] = filenames[0]
- data[comm_name]['sai1' ] = os.path.join( MAP_FOLDER , os.path.basename( filenames[0] ) + '.sai' )
+ data[comm_name]['sai1' ] = os.path.join( setup['map_folder'] , os.path.basename( filenames[0] ) + '.sai' )
data[comm_name]['fastq1cmd' ] = data[comm_name]['fastq1']
data[comm_name]['fastq2' ] = filenames[1]
- data[comm_name]['sai2' ] = os.path.join( MAP_FOLDER , os.path.basename( filenames[1] ) + '.sai' )
+ data[comm_name]['sai2' ] = os.path.join( setup['map_folder'] , os.path.basename( filenames[1] ) + '.sai' )
data[comm_name]['fastq2cmd' ] = data[comm_name]['fastq2']
@@ -669,19 +718,24 @@ def processIllumina(data):
##########################
## GENERATE COMMANDS
##########################
-def run_bwa(data):
- db_name = os.path.join(DB_FOLDER , os.path.basename( CONTIG ).replace('/', '_'))
- bam_name = os.path.join(OUT_FOLDER, os.path.basename( CONTIG ).replace('/', '_'))
+def run_bwa(data, setup):
+ in_fasta_bn = os.path.basename( setup['in_fasta'] ).replace('/', '_')
+ db_name = os.path.join( setup['db_folder' ], in_fasta_bn )
+ bam_name = os.path.join( setup['out_folder'], in_fasta_bn )
+ pre_ok = os.path.join( setup['out_folder'], in_fasta_bn + '.ok' )
+
bwt_index = db_name + '.bwt'
+
cmds = []
bams = []
deps = []
+ oks = []
sum_process = 0
- index_data = { 'db_name': db_name, 'contig' : CONTIG, 'bwa_exe': BWA_EXE}
+ index_data = { 'db_name': db_name, 'contig' : setup['in_fasta'], 'bwa_exe': BWA_EXE}
CMD_BWA_INDEX = "%(bwa_exe)s index -p %(db_name)s -a is %(contig)s"
- addCommand( 'bwa indexing', CMD_BWA_INDEX, TARGET=bwt_index, SOURCE=CONTIG, DATA=index_data, DEPS=[CONTIG], PRECIOUS=True )
+ addCommand( 'bwa indexing', CMD_BWA_INDEX, TARGET=bwt_index, SOURCE=setup['in_fasta'], DATA=index_data, DEPS=[], PRECIOUS=True )
##########################
@@ -705,7 +759,7 @@ def run_bwa(data):
CMD_CONVERT = "%(samtools_exe)s view -S -b -q 30 -1 %(sam)s > %(TARGET)s" # TARGET = bam_tmp
#CMD_MV1 = "mv %(bam_tmp_rnd)s %(bam_tmp)s"
- CMD_SORT = "%(samtools_exe)s sort -m 53687091200 -o %(bam_tmp)s > %(TARGET)s"
+ CMD_SORT = "%(samtools_exe)s sort -m 53687091200 -f %(bam_tmp)s %(TARGET)s"
#CMD_MV2 = "mv %(bam_pfx_rnd_bam)s %(bam)s"
CMD_INDEX = '%(samtools_exe)s index %(bam)s'
CMD_COV = '%(samtools_exe)s depth %(bam)s > %(cov)s'
@@ -735,8 +789,8 @@ def run_bwa(data):
addCommand('coverage' , CMD_COV , TARGET = info['cov'] , SOURCE = info['bam'] , DATA = info, DEPS = [ info['bai'] ] )
addCommand('stats' , CMD_STATS , TARGET = info['stats'] , SOURCE = info['bam'] , DATA = info, DEPS = [ info['bai'] ] )
addCommand('flag' , CMD_FLAG , TARGET = info['flag'] , SOURCE = info['bam'] , DATA = info, DEPS = [ info['bai'] ] )
- addCommand('oking' , CMD_OK , TARGET = info['ok'] , SOURCE = [ info['bai'], info['cov'], info['stats'], info['flag'] ], DATA = info )
+ deps = [ info['bai'], info['cov'], info['stats'], info['flag'] ]
if FRC:
if 'gsize' in info:
@@ -754,23 +808,26 @@ def run_bwa(data):
else:
print " SKIPPING FRC %(frc)s. disabled\n" % info
+ addCommand('oking' , CMD_OK , TARGET = info['ok'] , SOURCE = deps, DATA = info )
- deps.append( info['ok'] )
+ oks.append(info['ok'])
bams.append( info['bam'] )
-
-
+ CMD_PRE_OK = touch
+ addCommand('preoking' , CMD_PRE_OK , TARGET = pre_ok , SOURCE = oks, DATA = data )
##########################
## IF NOT TO MERGE, STOPS HERE
##########################
+ setup['data'] = data
if not do_merge:
- print "FINISHED"
- if __name__ == 'SCons.Script':
- env.Alias( 'all', deps )
- return
+ print "FINISHED", setup['proj_name']
+ env.Alias( setup['proj_name'], pre_ok )
+ setup['bamdata'] = ""
+ setup['outfile'] = pre_ok
+ return [ pre_ok ]
@@ -854,7 +911,8 @@ def run_bwa(data):
CMD_MERGE_tmp = False
if (not os.path.exists( bam_data['bam'] )) or ( os.path.getsize(bam_data['bam']) == 0 ):
CMD_MERGE_tmp = True
- addCommand('merging' , CMD_MERGE , TARGET = bam_data['bam'] , SOURCE = bams , DATA = bam_data, DEPS = deps , USETMP=CMD_MERGE_tmp )
+
+ addCommand('merging' , CMD_MERGE , TARGET = bam_data['bam'] , SOURCE = bams , DATA = bam_data, DEPS = [ pre_ok ] , USETMP=CMD_MERGE_tmp )
#addCommand('moving' , CMD_MV1 , TARGET = bam_data['bam'] , SOURCE = bam_data['bamrnd1'], DATA = bam_data )
@@ -887,7 +945,11 @@ def run_bwa(data):
if __name__ == 'SCons.Script':
- env.Alias( 'all', [ bam_data['fok'] ] )
+ print "FINISHED", setup['proj_name']
+ env.Alias( setup['proj_name'], bam_data['fok'] )
+ setup['bamdata'] = bam_data
+ setup['outfile'] = bam_data['fok']
+ return [ bam_data['fok'] ]
diff --git a/sp1/sp.csv b/sp1/sp.csv
index a9db75d706ca0ca5915532a8d430b713ac52f295..f1dacf430d315280df1e55810be903e76f9fdfd3 100644
--- a/sp1/sp.csv
+++ b/sp1/sp.csv
@@ -1,13 +1,13 @@
#group name platform library name library size library size min library size max file names
-120512_I238_FCC0U42ACXX_L1_SZAXPI008753-79 illumina pe 500 450 550 /home/assembly/tomato150/reseq/raw/074/illumina/pairedend_500/120512_I238_FCC0U42ACXX_L1_SZAXPI008753-79_1.fq.gz.clean.dup.clean.gz;/home/assembly/tomato150/reseq/raw/074/illumina/pairedend_500/120512_I238_FCC0U42ACXX_L1_SZAXPI008753-79_2.fq.gz.clean.dup.clean.gz
-120524_I238_FCD1174ACXX_L4_SZAXPI009266-140 illumina pe 170 153 187 /home/assembly/tomato150/denovo/pennellii/raw/illumina/pairedend_170/120524_I238_FCD1174ACXX_L4_SZAXPI009266-140_2.fq.gz.clean.dup.clean.gz;/home/assembly/tomato150/denovo/pennellii/raw/illumina/pairedend_170/120524_I238_FCD1174ACXX_L4_SZAXPI009266-140_1.fq.gz.clean.dup.clean.gz
-120530_I235_FCC0U06ACXX_L4_SZAXPI009266-140 illumina pe 170 153 187 /home/assembly/tomato150/denovo/pennellii/raw/illumina/pairedend_170/120530_I235_FCC0U06ACXX_L4_SZAXPI009266-140_1.fq.gz.clean.dup.clean.gz;/home/assembly/tomato150/denovo/pennellii/raw/illumina/pairedend_170/120530_I235_FCC0U06ACXX_L4_SZAXPI009266-140_2.fq.gz.clean.dup.clean.gz
-120530_I235_FCD117YACXX_L1_SZAXPI009266-140 illumina pe 170 153 187 /home/assembly/tomato150/denovo/pennellii/raw/illumina/pairedend_170/120530_I235_FCD117YACXX_L1_SZAXPI009266-140_2.fq.gz.clean.dup.clean.gz;/home/assembly/tomato150/denovo/pennellii/raw/illumina/pairedend_170/120530_I235_FCD117YACXX_L1_SZAXPI009266-140_1.fq.gz.clean.dup.clean.gz
-120530_I235_FCD117YACXX_L8_TOMxitDGLDWAAPEI-89 illumina mp 2000 1800 2200 /home/assembly/tomato150/denovo/pennellii/raw/illumina/matepair_2000/120530_I235_FCD117YACXX_L8_TOMxitDGLDWAAPEI-89_2.fq.gz.clean.dup.clean.gz;/home/assembly/tomato150/denovo/pennellii/raw/illumina/matepair_2000/120530_I235_FCD117YACXX_L8_TOMxitDGLDWAAPEI-89_1.fq.gz.clean.dup.clean.gz
-120615_I248_FCD14ADACXX_L8_TOMxitDGLDWAAPEI-89 illumina mp 2000 1800 2200 /home/assembly/tomato150/denovo/pennellii/raw/illumina/matepair_2000/120615_I248_FCD14ADACXX_L8_TOMxitDGLDWAAPEI-89_1.fq.gz.clean.dup.clean.gz;/home/assembly/tomato150/denovo/pennellii/raw/illumina/matepair_2000/120615_I248_FCD14ADACXX_L8_TOMxitDGLDWAAPEI-89_2.fq.gz.clean.dup.clean.gz
-120624_I232_FCC0U3HACXX_L6_TOMxitDGLDWAAPEI-89 illumina mp 2000 1800 2200 /home/assembly/tomato150/denovo/pennellii/raw/illumina/matepair_2000/120624_I232_FCC0U3HACXX_L6_TOMxitDGLDWAAPEI-89_1.fq.gz.clean.dup.clean.gz;/home/assembly/tomato150/denovo/pennellii/raw/illumina/matepair_2000/120624_I232_FCC0U3HACXX_L6_TOMxitDGLDWAAPEI-89_2.fq.gz.clean.dup.clean.gz
-120524_I238_FCD1174ACXX_L8_TOMxitDGLDWAAPEI-89 illumina mp 2000 1800 2200 /home/assembly/tomato150/denovo/pennellii/raw/illumina/matepair_2000/120524_I238_FCD1174ACXX_L8_TOMxitDGLDWAAPEI-89_2.fq.gz.clean.dup.clean.gz;/home/assembly/tomato150/denovo/pennellii/raw/illumina/matepair_2000/120524_I238_FCD1174ACXX_L8_TOMxitDGLDWAAPEI-89_1.fq.gz.clean.dup.clean.gz
-120527_I247_FCC0U0NACXX_L8_TOMxitDGLDWAAPEI-89 illumina mp 2000 1800 2200 /home/assembly/tomato150/denovo/pennellii/raw/illumina/matepair_2000/120527_I247_FCC0U0NACXX_L8_TOMxitDGLDWAAPEI-89_2.fq.gz.clean.dup.clean.gz;/home/assembly/tomato150/denovo/pennellii/raw/illumina/matepair_2000/120527_I247_FCC0U0NACXX_L8_TOMxitDGLDWAAPEI-89_1.fq.gz.clean.dup.clean.gz
-20000.dedup 454 mp 20000 18000 22000 /home/assembly/tomato150/denovo/pennellii/filtered/454/20000/20000.dedup.sff
-8000.dedup 454 mp 8000 7200 8800 /home/assembly/tomato150/denovo/pennellii/filtered/454/8000/8000.dedup.sff
+120524_I238_FCD1174ACXX_L4_SZAXPI009266-140 illumina pe 170 153 187 /home/assembly/tomato150/denovo/pennellii/filtered/illumina/pairedend_170/120524_I238_FCD1174ACXX_L4_SZAXPI009266-140_2.fq.gz.clean.dup.clean.gz;/home/assembly/tomato150/denovo/pennellii/filtered/illumina/pairedend_170/120524_I238_FCD1174ACXX_L4_SZAXPI009266-140_1.fq.gz.clean.dup.clean.gz
+120530_I235_FCC0U06ACXX_L4_SZAXPI009266-140 illumina pe 170 153 187 /home/assembly/tomato150/denovo/pennellii/filtered/illumina/pairedend_170/120530_I235_FCC0U06ACXX_L4_SZAXPI009266-140_1.fq.gz.clean.dup.clean.gz;/home/assembly/tomato150/denovo/pennellii/filtered/illumina/pairedend_170/120530_I235_FCC0U06ACXX_L4_SZAXPI009266-140_2.fq.gz.clean.dup.clean.gz
+120530_I235_FCD117YACXX_L1_SZAXPI009266-140 illumina pe 170 153 187 /home/assembly/tomato150/denovo/pennellii/filtered/illumina/pairedend_170/120530_I235_FCD117YACXX_L1_SZAXPI009266-140_2.fq.gz.clean.dup.clean.gz;/home/assembly/tomato150/denovo/pennellii/filtered/illumina/pairedend_170/120530_I235_FCD117YACXX_L1_SZAXPI009266-140_1.fq.gz.clean.dup.clean.gz
+120512_I238_FCC0U42ACXX_L1_SZAXPI008753-79 illumina pe 500 450 550 /home/assembly/tomato150/reseq/filtered/074/illumina/pairedend_500/120512_I238_FCC0U42ACXX_L1_SZAXPI008753-79_1.fq.gz.clean.dup.clean.gz;/home/assembly/tomato150/reseq/filtered/074/illumina/pairedend_500/120512_I238_FCC0U42ACXX_L1_SZAXPI008753-79_2.fq.gz.clean.dup.clean.gz
+120530_I235_FCD117YACXX_L8_TOMxitDGLDWAAPEI-89 illumina mp 2000 1800 2200 /home/assembly/tomato150/denovo/pennellii/filtered/illumina/matepair_2000/120530_I235_FCD117YACXX_L8_TOMxitDGLDWAAPEI-89_2.fq.gz.clean.dup.clean.gz;/home/assembly/tomato150/denovo/pennellii/filtered/illumina/matepair_2000/120530_I235_FCD117YACXX_L8_TOMxitDGLDWAAPEI-89_1.fq.gz.clean.dup.clean.gz
+120615_I248_FCD14ADACXX_L8_TOMxitDGLDWAAPEI-89 illumina mp 2000 1800 2200 /home/assembly/tomato150/denovo/pennellii/filtered/illumina/matepair_2000/120615_I248_FCD14ADACXX_L8_TOMxitDGLDWAAPEI-89_1.fq.gz.clean.dup.clean.gz;/home/assembly/tomato150/denovo/pennellii/filtered/illumina/matepair_2000/120615_I248_FCD14ADACXX_L8_TOMxitDGLDWAAPEI-89_2.fq.gz.clean.dup.clean.gz
+120624_I232_FCC0U3HACXX_L6_TOMxitDGLDWAAPEI-89 illumina mp 2000 1800 2200 /home/assembly/tomato150/denovo/pennellii/filtered/illumina/matepair_2000/120624_I232_FCC0U3HACXX_L6_TOMxitDGLDWAAPEI-89_1.fq.gz.clean.dup.clean.gz;/home/assembly/tomato150/denovo/pennellii/filtered/illumina/matepair_2000/120624_I232_FCC0U3HACXX_L6_TOMxitDGLDWAAPEI-89_2.fq.gz.clean.dup.clean.gz
+120524_I238_FCD1174ACXX_L8_TOMxitDGLDWAAPEI-89 illumina mp 2000 1800 2200 /home/assembly/tomato150/denovo/pennellii/filtered/illumina/matepair_2000/120524_I238_FCD1174ACXX_L8_TOMxitDGLDWAAPEI-89_2.fq.gz.clean.dup.clean.gz;/home/assembly/tomato150/denovo/pennellii/filtered/illumina/matepair_2000/120524_I238_FCD1174ACXX_L8_TOMxitDGLDWAAPEI-89_1.fq.gz.clean.dup.clean.gz
+120527_I247_FCC0U0NACXX_L8_TOMxitDGLDWAAPEI-89 illumina mp 2000 1800 2200 /home/assembly/tomato150/denovo/pennellii/filtered/illumina/matepair_2000/120527_I247_FCC0U0NACXX_L8_TOMxitDGLDWAAPEI-89_2.fq.gz.clean.dup.clean.gz;/home/assembly/tomato150/denovo/pennellii/filtered/illumina/matepair_2000/120527_I247_FCC0U0NACXX_L8_TOMxitDGLDWAAPEI-89_1.fq.gz.clean.dup.clean.gz
+20000.dedup 454 mp 20000 18000 22000 /home/assembly/tomato150/denovo/pennellii/filtered/454/20000/20000.dedup.sff
+8000.dedup 454 mp 8000 7200 8800 /home/assembly/tomato150/denovo/pennellii/filtered/454/8000/8000.dedup.sff
3000.dedup 454 mp 3000 2700 3300 /home/assembly/tomato150/denovo/pennellii/filtered/454/3000/3000.dedup.sff