def main() -> None:
logging.basicConfig(level=logging.DEBUG,
format="%(asctime)s:%(levelname)s: %(message)s")
p = argparse.ArgumentParser()
p.add_argument('--cache_dir',
default=util.DEFAULT_CACHE_DIR,
help='Benchmark cache dir')
p.add_argument('--result_dir',
default=util.DEFAULT_RESULT_DIR,
help='Benchmark result dir')
args = p.parse_args()
util.s3_cache_files([
util.REFERENCE_DIR + '/gencode.v26.whole_genes.fa',
util.REFERENCE_DIR + '/all_pair_art_lod_gpair_merged.txt'
], args.cache_dir)
for sample in util.TITRATION_SAMPLES:
logging.info('Start benchmark %s', sample.name)
result_dir = args.result_dir + '/' + sample.name
try:
os.makedirs(result_dir, 0o755)
except:
logging.error("mkdir %s failed", result_dir)
if os.path.exists(result_dir + "/filtered.fa"):
logging.info("Skip %s", result_dir)
continue
util.s3_cache_files(util.expand_fastq_files(sample.paths),
args.cache_dir)
cached_r1 = ",".join([
args.cache_dir + '/' + os.path.basename(fq.r1)
for fq in sample.paths
])
cached_r2 = ",".join([
args.cache_dir + '/' + os.path.basename(fq.r2)
for fq in sample.paths
])
cached_ref = args.cache_dir + '/gencode.v26.whole_genes.fa'
cached_cosmic_fusion = args.cache_dir + '/all_pair_art_lod_gpair_merged.txt'
af4_args = [
str(util.af4_path()), f'-log_dir={result_dir}', f'-pprof=:12345',
f'-mutex-profile-rate=1000', f'-block-profile-rate=1000',
f'-r1={cached_r1}', f'-r2={cached_r2}',
f'-fasta-output={result_dir}/all.fa',
f'-filtered-output={result_dir}/filtered.fa',
f'-transcript={cached_ref}', f'-max-genes-per-kmer=2',
f'-max-proximity-distance=1000', f'-max-proximity-genes=5',
f'-unstranded-prep', f'-cosmic-fusion={cached_cosmic_fusion}'
]
util.check_call(af4_args)
logging.info('Finished benchmark %d: %s', sample.name)
logging.info("Runtime stats: %s", util.run_stats(Path(result_dir)))
for path in glob.glob(f'{args.cache_dir}/*rerun*'):
try:
os.remove(path)
except:
logging.error("failed to remove " + path)
def main() -> None:
logging.basicConfig(level=logging.DEBUG,
format="%(asctime)s:%(levelname)s: %(message)s")
p = argparse.ArgumentParser()
p.add_argument('--cache_dir',
default=util.DEFAULT_CACHE_DIR,
help='Benchmark cache dir')
p.add_argument('--result_dir',
default=util.DEFAULT_RESULT_DIR,
help='Benchmark result dir')
p.add_argument('--starfusion_data_dir',
default='/scratch-nvme/starfusion',
help='Directory for expanding starfusion plug-n-play files')
p.add_argument(
'--run',
action='append',
choices=['af4', 'starfusion'],
help='List of systems to run. If unset, run all the configured systems'
)
p.add_argument(
'--starfusion_plug_n_play_targz',
default=os.environ['HOME'] +
'/GRCh38_v27_CTAT_lib_Feb092018.plug-n-play.tar.gz',
help=
'Tar.gz file of starfusion plug-n-play file. https://github.com/STAR-Fusion/STAR-Fusion/wiki#data-resources-required'
)
p.add_argument(
'--starfusion_targz',
default=os.environ['HOME'] + '/STAR-Fusion-v1.5.0.FULL.tar.gz',
help=
'Tar.gz file of starfusion source package. https://github.com/STAR-Fusion/STAR-Fusion/wiki#data-resources-required'
)
args = p.parse_args()
if not args.run:
args.run = ['af4', 'starfusion']
for sample in util.RNA_SAMPLES:
fastq_files: List[str] = []
cached_file_pairs: List[util.FASTQPair] = []
for fp in sample.paths:
assert fp.r1.replace("R1", "R2") == fp.r2, fp.r2
fastq_files += [fp.r1, fp.r2]
cached_file_pairs.append(
util.FASTQPair(
r1=args.cache_dir + '/' + os.path.basename(fp.r1),
r2=args.cache_dir + '/' + os.path.basename(fp.r2)))
util.s3_cache_files(fastq_files, args.cache_dir)
if 'af4' in args.run:
run_af4(sample.name, cached_file_pairs, args)
if 'starfusion' in args.run:
run_starfusion(sample.name, cached_file_pairs, args)
def run_af4(
sample_name: str,
cached_file_pairs: List[util.FASTQPair],
cosmic_fusion_path: str,
args: Any,
):
ref_path = "s3://grail-publications/resources/gencode.v26.whole_genes.fa"
util.s3_cache_files([ref_path, cosmic_fusion_path], args.cache_dir)
cached_r1 = ",".join([
args.cache_dir + "/" + os.path.basename(fp.r1)
for fp in cached_file_pairs
])
cached_r2 = ",".join([
args.cache_dir + "/" + os.path.basename(fp.r2)
for fp in cached_file_pairs
])
for mode in ["denovo", "targeted"]:
result_dir = args.result_dir + "/" + os.path.basename(sample_name +
"-" + mode)
if os.path.exists(result_dir + "/filtered.fa"):
logging.info("Skipping benchmark: %s", result_dir)
continue
logging.info("Start af4 benchmark: %s", result_dir)
try:
os.makedirs(result_dir, 0o755)
except:
logging.error("mkdir %s failed", result_dir)
af4_args = [
str(util.af4_path()),
f"-log_dir={result_dir}",
f"-pprof=:12345",
f"-mutex-profile-rate=1000",
f"-block-profile-rate=1000",
f"-r1={cached_r1}",
f"-r2={cached_r2}",
f"-max-genes-per-kmer=2",
f"-max-proximity-distance=1000",
f"-max-proximity-genes=5",
f"-fasta-output={result_dir}/all.fa",
f"-filtered-output={result_dir}/filtered.fa",
f"-transcript={args.cache_dir}/gencode.v26.250padded_separate_jns_transcripts_parsed_no_mt_no_overlap_no_pary_no_versioned.fa",
]
if mode == "targeted":
af4_args.append(
f"-cosmic-fusion={args.cache_dir}/all_pair_art_lod_gpair_merged.txt"
)
util.check_call(af4_args)
logging.info("Finished benchmark: %s", result_dir)
logging.info("Runtime stats: %s", util.run_stats(Path(result_dir)))
def main() -> None:
logging.basicConfig(level=logging.DEBUG,
format="%(asctime)s:%(levelname)s: %(messge)s")
p = argparse.ArgumentParser()
p.add_argument(
"--starfusion_dir",
default="/scratch-nvme/xyang/result/",
help=
"Starfusion result dir, which then contains individual sample result",
)
p.add_argument("--cache_dir",
default="/scratch-nvme/xyang/cache/",
help="cache dir")
args = p.parse_args()
local_truth_gpair = f"{args.cache_dir}/liu_gpair.txt"
if not os.path.exists(local_truth_gpair):
s3_cache_files([REFERENCE_DIR + "/liu_gpair.txt"], args.cache_dir)
listOfFiles = list()
for (dirpath, dirnames, filenames) in os.walk(args.starfusion_dir):
listOfFiles += [os.path.join(dirpath, file) for file in filenames]
# get all relevant starfusion result files
filtered_results = [
f for f in listOfFiles
if f.find("star-fusion.fusion_predictions.abridged.tsv") != -1
]
# compare with truth set
true_gpairs = read_fusion_pair(local_truth_gpair)
npos = len(true_gpairs)
for f in filtered_results:
print(f)
results = read_fusion_pair(f)
npredict = len(results)
tp = len(true_gpairs.intersection(results))
fn = npos - tp
fp = npredict - tp
precision = tp / (tp + fp)
recall = tp / npos
f1 = 2 * precision * recall / (precision + recall)
f1 = "%.3f" % (f1)
print(f"tp={tp}, fn={fn}, fp={fp}, f1={f1}")
def main() -> None:
logging.basicConfig(
level=logging.DEBUG, format="%(asctime)s:%(levelname)s: %(message)s"
)
p = argparse.ArgumentParser()
p.add_argument(
"--cache_dir", default=util.DEFAULT_CACHE_DIR, help="Benchmark cache dir"
)
p.add_argument(
"--result_dir", default=util.DEFAULT_RESULT_DIR, help="Benchmark result dir"
)
p.add_argument(
"--starfusion_data_dir",
default="/scratch-nvme/starfusion",
help="Directory for expanding starfusion plug-n-play files",
)
p.add_argument(
"--starfusion_plug_n_play_targz",
default=os.environ["HOME"]
+ "/GRCh38_v27_CTAT_lib_Feb092018.plug-n-play.tar.gz",
help="Tar.gz file of starfusion plug-n-play file. https://github.com/STAR-Fusion/STAR-Fusion/wiki#data-resources-required",
)
p.add_argument(
"--starfusion_targz",
default=os.environ["HOME"] + "/STAR-Fusion-v1.5.0.FULL.tar.gz",
help="Tar.gz file of starfusion source package. https://github.com/STAR-Fusion/STAR-Fusion/wiki#data-resources-required",
)
args = p.parse_args()
for sample in util.SIMULATED_SAMPLES:
util.s3_cache_files([sample.path.r1, sample.path.r2], args.cache_dir)
fastq_files: List[str] = []
cached_file_pairs: List[util.FASTQPair] = []
fastq_files += [sample.path.r1, sample.path.r2]
cached_file_pairs.append(
util.FASTQPair(
r1=args.cache_dir + "/" + os.path.basename(sample.path.r1),
r2=args.cache_dir + "/" + os.path.basename(sample.path.r2),
)
)
print(cached_file_pairs)
sample_name = str(sample.n) + "_" + str(sample.coverage)
run_starfusion(sample_name, cached_file_pairs, args)
def run_af4(sample_name: str, cached_file_pairs: List[util.FASTQPair],
args: Any):
ref_path = "s3://grail-publications/resources/gencode.v26.whole_genes.fa"
cosmic_fusion_path = "s3://grail-publications/resources/all_pair_art_lod_gpair_merged.txt"
util.s3_cache_files([ref_path, cosmic_fusion_path], args.cache_dir)
cached_r1 = ",".join([
args.cache_dir + '/' + os.path.basename(fp.r1)
for fp in cached_file_pairs
])
cached_r2 = ",".join([
args.cache_dir + '/' + os.path.basename(fp.r2)
for fp in cached_file_pairs
])
for mode in ['denovo', 'targeted']:
result_dir = args.result_dir + '/' + os.path.basename(sample_name +
'-' + mode)
if os.path.exists(result_dir + "/filtered.fa"):
logging.info('Skipping benchmark: %s', result_dir)
continue
logging.info('Start af4 benchmark: %s', result_dir)
try:
os.makedirs(result_dir, 0o755)
except:
logging.error("mkdir %s failed", result_dir)
af4_args = [
str(util.af4_path()), f'-log_dir={result_dir}', f'-pprof=:12345',
f'-mutex-profile-rate=1000', f'-block-profile-rate=1000',
f'-umi-in-read', f'-r1={cached_r1}', f'-r2={cached_r2}',
f'-fasta-output={result_dir}/all.fa',
f'-filtered-output={result_dir}/filtered.fa',
f'-transcript={args.cache_dir}/gencode.v26.250padded_separate_jns_transcripts_parsed_no_mt_no_overlap_no_pary_no_versioned.fa'
]
if mode == 'targeted':
af4_args.append(
f'-cosmic-fusion={args.cache_dir}/all_pair_art_lod_gpair_merged.txt'
)
util.check_call(af4_args)
logging.info('Finished benchmark: %s', result_dir)
logging.info("Runtime stats: %s", util.run_stats(Path(result_dir)))
def main() -> None:
logging.basicConfig(level=logging.DEBUG,
format="%(asctime)s:%(levelname)s: %(message)s")
p = argparse.ArgumentParser()
p.add_argument("--cache_dir",
default=util.DEFAULT_CACHE_DIR,
help="Benchmark cache dir")
p.add_argument("--result_dir",
default=util.DEFAULT_RESULT_DIR,
help="Benchmark result dir")
p.add_argument(
"--starfusion_data_dir",
default="/scratch-nvme/starfusion",
help="Directory for expanding starfusion plug-n-play files",
)
p.add_argument(
"--run",
action="append",
choices=["af4", "starfusion"],
help="List of systems to run. If unset, run all the configured systems",
)
p.add_argument(
"--starfusion_plug_n_play_targz",
default=os.environ["HOME"] +
"/GRCh38_v27_CTAT_lib_Feb092018.plug-n-play.tar.gz",
help=
"Tar.gz file of starfusion plug-n-play file. https://github.com/STAR-Fusion/STAR-Fusion/wiki#data-resources-required",
)
p.add_argument(
"--starfusion_targz",
default=os.environ["HOME"] + "/STAR-Fusion-v1.5.0.FULL.tar.gz",
help=
"Tar.gz file of starfusion source package. https://github.com/STAR-Fusion/STAR-Fusion/wiki#data-resources-required",
)
p.add_argument(
"--brca_data_dir",
default="/scratch-nvme/xyang/brca_rnaseq_data",
help="BT474, KPL4, MCF7, SKBR3 Breast cancer data directory",
)
args = p.parse_args()
if not args.run:
args.run = ["af4", "starfusion"]
## brca rna-seq for af4
brca_samples = [
os.path.join(args.brca_data_dir, s)
for s in ["BT474", "KPL4", "MCF7", "SKBR3"]
]
for s in brca_samples:
if not os.path.exists(os.path.join(args.brca_data_dir, s)):
util.check_call([
"download_brca_data.py",
"--odir",
"/scratch-nvme/xyang/brca_rnaseq_data",
])
cosmic_fusion_path = (
"s3://grail-publications/2019-ISMB/references/all_art_lod_brca.txt")
for sample in brca_samples:
r1s: List[str] = []
for fq in os.listdir(sample):
if "_1" in fq:
r1s.append(os.path.join(sample, fq))
cached_file_pairs: List[util.FASTQPair] = []
for r1 in r1s:
assert os.path.exists(r1.replace("_1", "_2"))
cached_file_pairs.append(
util.FASTQPair(r1=r1, r2=r1.replace("_1", "_2")))
print(os.path.basename(sample))
print(cached_file_pairs)
run_af4(os.path.basename(sample), cached_file_pairs,
cosmic_fusion_path, args)
## cfrna for af4 and starfusion
cosmic_fusion_path = (
"s3://grail-publications/2019-ISMB/references/all_pair_art_lod_gpair_merged.txt"
)
for sample in util.RNA_SAMPLES:
fastq_files: List[str] = []
cached_file_pairs: List[util.FASTQPair] = []
for fp in sample.paths:
assert fp.r1.replace("R1", "R2") == fp.r2, fp.r2
fastq_files += [fp.r1, fp.r2]
cached_file_pairs.append(
util.FASTQPair(
r1=args.cache_dir + "/" + os.path.basename(fp.r1),
r2=args.cache_dir + "/" + os.path.basename(fp.r2),
))
util.s3_cache_files(fastq_files, args.cache_dir)
if "af4" in args.run:
run_af4(sample.name, cached_file_pairs, cosmic_fusion_path, args)
if "starfusion" in args.run:
run_starfusion(sample.name, cached_file_pairs, args)
def main() -> None:
logging.basicConfig(level=logging.DEBUG,
format="%(asctime)s:%(levelname)s: %(message)s")
p = argparse.ArgumentParser()
p.add_argument('--cache_dir',
default=util.DEFAULT_CACHE_DIR,
help='Benchmark cache dir')
p.add_argument('--result_dir',
default=util.DEFAULT_RESULT_DIR,
help='Benchmark result dir')
p.add_argument(
'--rerun_af4',
action='store_true',
help='Always run AF4 even if the result file already exists')
p.add_argument(
'--recache_files',
action='store_true',
help=
'Always copy benchmark data files, even if they already exist locally.'
)
args = p.parse_args()
util.s3_cache_files([
util.REFERENCE_DIR +
'/gencode.v26.250padded_separate_jns_transcripts_parsed_no_mt_no_overlap_no_pary_no_versioned.fa',
util.REFERENCE_DIR + '/all_pair_art_lod_gpair_merged.txt',
util.REFERENCE_DIR + '/liu_gpair.txt'
], args.cache_dir)
for mode in ['denovo', 'targeted']:
for sample in util.SIMULATED_SAMPLES:
util.s3_cache_files([sample.path.r1, sample.path.r2],
args.cache_dir)
result_dir = f'{args.result_dir}/synthetic-{mode}-{sample.n}-{sample.coverage}'
try:
os.makedirs(result_dir, 0o755)
except:
logging.error("mkdir %s failed", result_dir)
if not os.path.exists(
f'{result_dir}/filtered.fa') or args.rerun_af4:
logging.info('running benchmark in %s', result_dir)
af4_args = [
str(util.af4_path()), f'-log_dir={result_dir}',
f'-r1={args.cache_dir}/{sample.path.r1}',
f'-r2={args.cache_dir}/{sample.path.r2}',
f'-fasta-output={result_dir}/all.fa',
f'-filtered-output={result_dir}/filtered.fa',
'-transcript=' + args.cache_dir +
'/gencode.v26.250padded_separate_jns_transcripts_parsed_no_mt_no_overlap_no_pary_no_versioned.fa'
]
if mode == 'targeted':
af4_args.append('-cosmic-fusion=' + args.cache_dir +
'/all_pair_art_lod_gpair_merged.txt')
util.check_call(af4_args)
logging.info("Runtime stats: %s",
util.run_stats(Path(result_dir)))
stats = TargetedFusionStats(
Path(f'{args.cache_dir}/liu_gpair.txt'),
Path(f'{result_dir}/filtered.fa'))
s = stats.stats()
tp = "%d" % (s.tp, )
fp = "%d" % (s.fp, )
fn = "%d" % (s.fn, )
print(
f'{mode} & {sample.n} & {sample.coverage} & {tp} & {fp} & {fn}\\\\'
)
def main() -> None:
logging.basicConfig(level=logging.DEBUG,
format="%(asctime)s:%(levelname)s: %(message)s")
p = argparse.ArgumentParser()
p.add_argument("--cache_dir",
default=util.DEFAULT_CACHE_DIR,
help="Benchmark cache dir")
p.add_argument("--result_dir",
default=util.DEFAULT_RESULT_DIR,
help="Benchmark result dir")
p.add_argument(
"--rerun_af4",
action="store_true",
help="Always run AF4 even if the result file already exists",
)
p.add_argument(
"--recache_files",
action="store_true",
help=
"Always copy benchmark data files, even if they already exist locally.",
)
args = p.parse_args()
util.s3_cache_files(
[
util.REFERENCE_DIR +
"/gencode.v26.250padded_separate_jns_transcripts_parsed_no_mt_no_overlap_no_pary_no_versioned.fa",
util.REFERENCE_DIR + "/all_pair_art_lod_gpair_merged.txt",
util.REFERENCE_DIR + "/liu_gpair.txt",
],
args.cache_dir,
)
for mode in ["denovo", "targeted"]:
for sample in util.SIMULATED_SAMPLES:
util.s3_cache_files([sample.path.r1, sample.path.r2],
args.cache_dir)
result_dir = (
f"{args.result_dir}/synthetic-{mode}-{sample.n}-{sample.coverage}"
)
try:
os.makedirs(result_dir, 0o755)
except:
logging.error("mkdir %s failed", result_dir)
if not os.path.exists(
f"{result_dir}/filtered.fa") or args.rerun_af4:
logging.info("running benchmark in %s", result_dir)
af4_args = [
str(util.af4_path()),
f"-log_dir={result_dir}",
f"-r1={args.cache_dir}/{sample.path.r1}",
f"-r2={args.cache_dir}/{sample.path.r2}",
f"-fasta-output={result_dir}/all.fa",
f"-filtered-output={result_dir}/filtered.fa",
f"-max-genes-per-kmer=2",
f"-max-proximity-distance=1000",
f"-max-proximity-genes=5",
"-transcript=" + args.cache_dir +
"/gencode.v26.250padded_separate_jns_transcripts_parsed_no_mt_no_overlap_no_pary_no_versioned.fa",
]
if mode == "targeted":
af4_args.append("-cosmic-fusion=" + args.cache_dir +
"/all_pair_art_lod_gpair_merged.txt")
util.check_call(af4_args)
logging.info("Runtime stats: %s",
util.run_stats(Path(result_dir)))
stats = TargetedFusionStats(
Path(f"{args.cache_dir}/liu_gpair.txt"),
Path(f"{result_dir}/filtered.fa"),
)
s = stats.stats()
tp = "%d" % (s.tp, )
fp = "%d" % (s.fp, )
fn = "%d" % (s.fn, )
print(
f"{mode} & {sample.n} & {sample.coverage} & {tp} & {fp} & {fn}\\\\"
)
