def merge(proc_dir, job1, jobn, n, stride):
pipeline = Plumber()
map_jobn = NuMap(worker_num=jobn,
ordered=False,
stride=stride,
buffer=10000)
p1 = Piper(Worker(merge_star, (proc_dir, n)), parallel=map_jobn)
p2 = Piper(Worker(npasser), parallel=map_jobn)
pipeline.add_pipe((p1, p2))
return pipeline
def csort(proc_dir, job1, jobn, memn, n, stride):
csort_mem = "%sG" % memn
pipeline = Plumber()
map_jobn = NuMap(worker_num=jobn,
ordered=False,
stride=stride,
buffer=10000)
p1 = Piper(Worker(sambamba_csort, (proc_dir, n, csort_mem)),
parallel=map_jobn)
p2 = Piper(Worker(npasser), parallel=map_jobn)
pipeline.add_pipe((p1, p2))
return pipeline
# >>> arr = stamper([arr])
# Step 3 (representing computational resources)
# creates a resource that allows to utilize all local processors
local_computer = NuMap()
# Step 4 (creating processing nodes)
# this attaches a single computational resource to the two processing nodes
# the stamper_node will be tracked i.e. it will store the results of computation
# in memory.
cleaner_node = Piper(cleaner, parallel=local_computer)
stamper_node = Piper(stamper, parallel=local_computer, track=True)
# Step 5 (constructing a workflow graph)
# we construct a workflow graph add the two processing nodes and define the
# connection between them.
workflow = Plumber()
workflow.add_pipe((cleaner_node, stamper_node))
# Step 6 (execute the workflow)
# this starts the workflow, processes data in the "background" and waits
# until all data-items have been processed.
workflow.start([['AGA.TA', 'TG..AA']])
workflow.run()
workflow.wait()
results = workflow.stats['pipers_tracked'][stamper_node][0]
for seq in results.values():
print "Object \"%s\" has time stamp: %s " % (seq, seq.meta['timestamp'])
def l33t(inbox):
word = inbox[0]
return word.replace('e', '3').replace('o', '0')
def l33ter(inbox):
word = inbox[0]
return word.replace('l', '1')
# execution endgine
numap = NuMap()
# function nodes
l33t_piper = Piper(l33t, parallel=numap)
l33ter_piper = Piper(l33ter, parallel=numap, track=True)
# topology
pipeline = Plumber()
pipeline.add_pipe((l33t_piper, l33ter_piper))
end = pipeline.get_outputs()[0]
# runtime
pipeline.start([['hello', 'world']])
pipeline.run()
pipeline.wait()
print pipeline.stats['pipers_tracked'][end]
assert [{0: 'h3110', 1: 'w0r1d'}] == pipeline.stats['pipers_tracked'][end]
from papy.core import Plumber, Piper
def l33t(inbox):
word = inbox[0]
return word.replace('e', '3').replace('o', '0')
def l33ter(inbox):
word = inbox[0]
return word.replace('l', '1')
# execution endgine
numap = NuMap()
# function nodes
l33t_piper = Piper(l33t, parallel=numap)
l33ter_piper = Piper(l33ter, parallel=numap, track=True)
# topology
pipeline = Plumber()
pipeline.add_pipe((l33t_piper, l33ter_piper))
end = pipeline.get_outputs()[0]
# runtime
pipeline.start([['hello', 'world']])
pipeline.run()
pipeline.wait()
print pipeline.stats['pipers_tracked'][end]
assert [{0: 'h3110', 1: 'w0r1d'}] == pipeline.stats['pipers_tracked'][end]
# wraps timestamp
stamper = Worker(timestamp)
# >>> arr = stamper([arr])
# Step 3 (representing computational resources)
# creates a resource that allows to utilize all local processors
local_computer = NuMap()
# Step 4 (creating processing nodes)
# this attaches a single computational resource to the two processing nodes
# the stamper_node will be tracked i.e. it will store the results of computation
# in memory.
cleaner_node = Piper(cleaner, parallel=local_computer)
stamper_node = Piper(stamper, parallel=local_computer, track=True)
# Step 5 (constructing a workflow graph)
# we construct a workflow graph add the two processing nodes and define the
# connection between them.
workflow = Plumber()
workflow.add_pipe((cleaner_node, stamper_node))
# Step 6 (execute the workflow)
# this starts the workflow, processes data in the "background" and waits
# until all data-items have been processed.
workflow.start([['AGA.TA', 'TG..AA']])
workflow.run()
workflow.wait()
results = workflow.stats['pipers_tracked'][stamper_node][0]
for seq in results.values():
print "Object \"%s\" has time stamp: %s " % (seq, seq.meta['timestamp'])
def final(proc_dir, job1, jobn, n, stride, full_model, genome_idx, genome_seq,
skip_mixcr, skip_cover):
## bamstat
bamstat_cfg = {
"id": "bamstat",
"evaluator": EVALUATOR,
"preamble": PREAMBLE,
"dir": proc_dir,
"executable": "bash",
"script": "%s/%s" % (WORK_DIR, "work_bamstat.sh"),
"in": ("alig_csort", ),
"out": (
("main", "log"),
("idxstat", "txt"),
("flgstat", "txt"),
),
"params": {}
}
virus_cfg = {
"id": "virus",
"evaluator": EVALUATOR,
"preamble": PREAMBLE,
"dir": proc_dir,
"executable": "bash",
"script": "%s/%s" % (WORK_DIR, "work_virus.sh"),
"in": ("alig_csort", ),
"out": (
("main", "log"),
("call", "txt"),
),
"params": {}
}
gzip_cfg = {
"id": "gzip",
"evaluator": EVALUATOR,
"preamble": PREAMBLE,
"dir": proc_dir,
"executable": "bash",
"script": "%s/%s" % (WORK_DIR, "work_gzip.sh"),
"in": (
"sj",
"junc_se",
"junc_pe",
),
"out": (("main", "log"), ),
"params": {}
}
cram_cfg = {
"id":
"pack",
"evaluator":
EVALUATOR,
"preamble":
PREAMBLE,
"dir":
proc_dir,
"executable":
"bash",
"script":
"%s/%s" % (WORK_DIR, "work_cram.sh"),
"in": (
"alig_csort",
"chim_pe_csort",
"chim_se_csort",
),
"out": (
("main", "log"),
("alig_csort", "cram"),
("chim_pe_csort", "cram"),
("chim_se_csort", "cram"),
),
"params": {
"genome": genome_seq,
"qbin": "2,10,20,25,30,35,40,42"
}
}
cover_cfg = {
"id":
"cover",
"evaluator":
EVALUATOR,
"preamble":
PREAMBLE,
"dir":
proc_dir,
"executable":
"python2",
"script":
"%s/%s" % (WORK_DIR, "work_coverage.py"),
"in": (
"alig_csort",
"chim_pe_csort",
"chim_se_csort",
),
"out": (
("main", "log"),
("alig_csort", "bw"),
("chim_pe_csort", "bw"),
("chim_se_csort", "bw"),
),
"params": {
"chrom_length": genome_idx / "chrNameLength.txt"
}
}
mixcr_cfg = {
"id":
"mixcr",
"evaluator":
EVALUATOR,
"preamble":
PREAMBLE,
"dir":
proc_dir,
"executable":
"bash",
"script":
"%s/%s" % (WORK_DIR, "work_mixcr.sh"),
"in": (
"alig_csort",
"unmap_1",
"unmap_2",
),
"out": (
("main", "log"),
("aln.rep", "txt"),
("fix1.rep", "txt"),
("fix2.rep", "txt"),
("ext.rep", "txt"),
("asm.rep", "txt"),
("index", "bin"),
("alig", "vdjca"),
("clone", "clns"),
("dir", None),
),
"params": {
"p": n,
"TRA": "chr14:21543538-22556037",
"TRB": "chr7:142290000-142820000",
"TRG": "chr7:38237000-38382000",
"IGK": "chr2:88789991-90313268",
"IGH": "chr14:105580000-106880000",
"IGL": "chr22:22020000-22927000",
}
}
## topology
pipeline = Plumber()
map_job1 = NuMap(worker_num=job1,
ordered=False,
stride=stride,
buffer=10000)
map_jobn = NuMap(worker_num=jobn,
ordered=False,
stride=stride,
buffer=10000)
p1 = Piper(Worker(ipasser), parallel=map_job1)
p2b = Piper(Worker(script, (cram_cfg, )), parallel=map_job1)
p2c = Piper(Worker(script, (gzip_cfg, )), parallel=map_job1)
p2d = Piper(Worker(script, (bamstat_cfg, )), parallel=map_job1)
p2e = Piper(Worker(script, (virus_cfg, )), parallel=map_job1)
p3 = Piper(Worker(npasser), parallel=map_job1)
pipeline.add_pipe((p1, p2b, p3))
pipeline.add_pipe((p1, p2c, p3))
pipeline.add_pipe((p1, p2d, p3))
pipeline.add_pipe((p1, p2e, p3))
if not skip_mixcr:
p2f = Piper(Worker(script, (mixcr_cfg, )), parallel=map_jobn)
pipeline.add_pipe((p1, p2f, p3))
if not skip_cover:
p2g = Piper(Worker(script, (cover_cfg, )), parallel=map_job1)
pipeline.add_pipe((p1, p2g, p3))
return pipeline
def quant(proc_dir, job1, jobn, n, stride, unstranded, prot_model, full_model,
linc_model):
## gene counting
prot_cfg = {
"id": "prot",
"evaluator": EVALUATOR,
"preamble": PREAMBLE,
"dir": proc_dir,
"executable": "python2",
"script": "%s/work_featc.py" % WORK_DIR,
"in": ("alig_nsort", ),
"out": (("main", "log"), ("cts", "cts"), ("tmp", None)),
"params": {
"paired_end": True,
"stranded": "0", # always unstranded
"duplicates": "", # count duplicates
"gtf": prot_model,
"n": n,
"xargs": ""
}
}
full_cfg = {
"id": "full",
"evaluator": EVALUATOR,
"preamble": PREAMBLE,
"dir": proc_dir,
"executable": "python2",
"script": "%s/work_featc.py" % WORK_DIR,
"in": ("alig_nsort", ),
"out": (("main", "log"), ("cts", "cts"), ("tmp", None)),
"params": {
"paired_end": True,
"stranded": "0"
if unstranded else "1", # first read is on the transcript strand
"duplicates": "", # count duplicates
"gtf": full_model,
"n": n,
"xargs": "",
}
}
linc_cfg = {
"id": "linc",
"evaluator": EVALUATOR,
"preamble": PREAMBLE,
"dir": proc_dir,
"executable": "python2",
"script": "%s/work_featc.py" % WORK_DIR,
"in": ("alig_nsort", ),
"out": (("main", "log"), ("cts", "cts"), ("tmp", None)),
"params": {
"paired_end": True,
"stranded": "0"
if unstranded else "1", # first read is on the transcript strand
"duplicates": "", # count duplicates
"gtf": linc_model,
"n": n,
"xargs": ""
}
}
cfgs = [prot_cfg, full_cfg, linc_cfg]
## topology
pipeline = Plumber()
map_job1 = NuMap(worker_num=job1,
ordered=False,
stride=stride,
buffer=10000)
map_job4 = NuMap(worker_num=job1 / 4,
ordered=False,
stride=stride,
buffer=10000)
p1 = Piper(Worker(ipasser), parallel=map_job1)
p2 = Piper(Worker(npasser), parallel=map_job1)
for cfg in cfgs:
p = Piper(Worker(script, (cfg, )), parallel=map_job4)
pipeline.add_pipe((p1, p, p2))
return pipeline
def align(proc_dir, job1, jobn, memn, stride, n, unstranded, genome_idx,
full_model, genome_seq, rrna_seq, merge_mode, star_mem,
alig_star_params, chim_star_params, prune, keep_fastq):
cutmrg_script = "work_bbcutmrg_pe.sh" if merge_mode == "bb" else "work_cutmrg_pe.sh"
cutmrg_cfg = {
"id":
"cutmrg",
"evaluator":
EVALUATOR,
"preamble":
PREAMBLE,
"dir":
proc_dir,
"executable":
"bash",
"script":
"%s/%s" % (WORK_DIR, cutmrg_script),
"in": (
"fastq1",
"fastq2",
),
"out": (
("main", "log"),
("cut", "log"),
("mrg", "log"),
("cutfq1", "fq"),
("cutfq2", "fq"),
("mrgfq1", "fq"),
("mrgfq2", "fq"),
("mrgfq3", "fq"),
("isize", "txt"),
("stats", "txt"),
("dir", None),
),
"params": {
"min_len": 25,
"cutxargs": "k=31 qskip=3 rieb=t tbo=t tpe=t",
"mrgxargs":
"k=31 prefilter=2 minoverlap=10 extend2=20 iterations=5",
"seq1": "AGATCGGAAGAGCACACGTCTGAACTCCAGTCAC",
"seq2":
"AGATCGGAAGAGCGTCGTGTAGGGAAAGAGTGTAGATCTCGGTGGTCGCCGTATCATT",
"rrna": rrna_seq,
"prune": prune,
"xmx": memn,
"p": n,
}
}
star_alig_params = {
"keep_fastq": keep_fastq,
"genomeDir": genome_idx,
"runThreadN": n,
"genomeLoad": star_mem,
## spliced alignment
"outFilterType": "BySJout",
"outSAMstrandField": "intronMotif" if unstranded else "None",
"alignSJoverhangMin": 8,
"alignSJDBoverhangMin": 3,
"scoreGenomicLengthLog2scale": 0,
"alignIntronMin": 20,
"alignIntronMax": 1000000,
"alignMatesGapMax": 1000000,
}
star_alig_params.update(alig_star_params)
star_chim_params = {
"keep_fastq": keep_fastq,
"genomeDir": genome_idx,
"runThreadN": n,
"genomeLoad": star_mem,
"outFilterType": "Normal",
## chimeric alignment
"alignIntronMax": 150000,
"alignMatesGapMax": 150000,
"chimSegmentMin": 10,
"chimJunctionOverhangMin": 1,
"chimScoreSeparation": 0,
"chimScoreJunctionNonGTAG": 0,
"chimScoreDropMax": 1000,
"chimScoreMin": 1,
}
star_chim_params.update(chim_star_params)
star_alig_cfg = {
"id": "alig",
"evaluator": EVALUATOR,
"preamble": PREAMBLE,
"dir": proc_dir,
"executable": "python2",
"script": "%s/%s" % (WORK_DIR, "work_star_alig_pe.py"),
"in": (
"cutfq1",
"cutfq2",
),
"out": (
("main", "log"),
("dir", None),
),
"params": star_alig_params
}
star_chim_cfg = {
"id": "chim",
"evaluator": EVALUATOR,
"preamble": PREAMBLE,
"dir": proc_dir,
"executable": "python2",
"script": "%s/%s" % (WORK_DIR, "work_star_chim_pe.py"),
"in": (
"mrgfq1",
"mrgfq2",
"mrgfq3",
),
"out": (
("main", "log"),
("dir", None),
),
"params": star_chim_params
}
pipeline = Plumber()
map_job1 = NuMap(worker_num=job1,
ordered=False,
stride=stride,
buffer=10000)
map_jobn = NuMap(worker_num=jobn,
ordered=False,
stride=stride,
buffer=10000)
p1 = Piper(Worker(link_fq, (proc_dir, )), parallel=map_job1)
p2 = Piper(Worker(script, (cutmrg_cfg, )),
parallel=(map_jobn if merge_mode == "bb" else map_job1))
p3a = Piper(Worker(script, (star_alig_cfg, )), parallel=map_jobn)
p3b = Piper(Worker(script, (star_chim_cfg, )), parallel=map_jobn)
p4a = Piper(Worker(move_alig_star, (proc_dir, )), parallel=map_job1)
p4b = Piper(Worker(move_chim_star, (proc_dir, )), parallel=map_job1)
p5 = Piper(Worker(npasser), parallel=map_jobn)
pipeline.add_pipe((p1, p2, p3a, p4a, p5))
pipeline.add_pipe((p1, p2, p3b, p4b, p5))
return pipeline
def preqc(proc_dir, job1, jobn, n, stride):
kmer_cfg = {
"id": "kmer",
"evaluator": EVALUATOR,
"preamble": PREAMBLE,
"dir": proc_dir,
"executable": "bash",
"script": "%s/%s" % (WORK_DIR, "work_kmer.sh"),
"in": (
"fastq1",
"fastq2",
),
"out": (
("main", "log"),
("kmer1", "txt"),
("kmer2", "txt"),
),
"params": {
"k": 6,
}
}
fastqc_cfg = {
"id": "fastqc",
"evaluator": EVALUATOR,
"preamble": PREAMBLE,
"dir": proc_dir,
"executable": "bash",
"script": "%s/%s" % (WORK_DIR, "work_fastqc.sh"),
"in": (
"fastq1",
"fastq2",
),
"out": (
("main", "log"),
("report", None),
),
"params": {}
}
map_job1 = NuMap(worker_num=job1,
ordered=False,
stride=stride,
buffer=10000)
map_jobn = NuMap(worker_num=jobn,
ordered=False,
stride=stride,
buffer=10000)
p1 = Piper(Worker(sample_fq, (
proc_dir,
1000000,
)), parallel=map_job1)
p2 = Piper(Worker(script, (kmer_cfg, )), parallel=map_job1)
p3 = Piper(Worker(script, (fastqc_cfg, )), parallel=map_job1)
p4 = Piper(Worker(npasser), parallel=map_job1)
# topology
pipeline = Plumber()
pipeline.add_pipe((p1, p2, p4))
pipeline.add_pipe((p1, p3, p4))
return pipeline
