def run_dataset(self, case, algorithm):
bp = "test/{}/".format(case)
test_file = bp + "test.spats.tmp"
try:
shutil.copyfile(bp + "ds.spats", test_file)
db = PairDB(test_file)
s = Spats()
db.load_run(s.run)
if not s.run.cotrans and algorithm == "native":
return
s.run.writeback_results = True
s.run.result_set_name = "test"
s.run.algorithm = algorithm
s.run.quiet = True
s.loadTargets(db)
if not s._processor.exists():
# just ignore the native test if it's not available
self.assertEqual("native", algorithm)
return
s.process_pair_db(db, batch_size = 1024) # small batch_size just to exercise multiprocessing code
msg = None
count = 0
for res in db.differing_results("test", "test_validation"):
msg = str([str(x) for x in res])
count += 1
self.assertEqual(0, count, "{} differing results: {} / {} \n{}".format(count, case, algorithm, msg))
finally:
if os.path.exists(test_file):
os.remove(test_file)
def tags():
bp = "/Users/jbrink/mos/tasks/1RwIBa/tmp/datasets/cotrans/"
from spats_shape_seq.db import PairDB
pair_db = PairDB(bp + "db/pairs.db")
if True:
print "Parsing to db..."
pair_db.wipe()
pair_db.add_targets_table(bp + "cotrans_single.fa")
pair_db.parse(bp + "data/EJS_6_F_10mM_NaF_Rep1_GCCAAT_R1.fastq",
bp + "data/EJS_6_F_10mM_NaF_Rep1_GCCAAT_R2.fastq",
sample_size=100000)
from spats_shape_seq import Spats
from spats_shape_seq.tag import TagProcessor
from spats_shape_seq.util import reverse_complement
s = Spats()
s.run._processor_class = TagProcessor
s.run.writeback_results = True
s.run.result_set_name = "tags"
s.run.num_workers = 1
s.run.cotrans = True
s.run.cotrans_linker = 'CTGACTCGGGCACCAAGGAC'
s.loadTargets(pair_db)
s.run.allow_indeterminate = True
s.run.allowed_target_errors = 2
s.run.allowed_adapter_errors = 2
p = s._processor
for target in pair_db.targets():
p.addTagTarget(target[0], target[1])
p.addTagTarget(target[0] + "_rc", reverse_complement(target[1]))
p.addTagTarget("adapter_t_rc", reverse_complement(s.run.adapter_t))
p.addTagTarget("adapter_b", s.run.adapter_b)
if s.run.cotrans:
p.addTagTarget("linker_cotrans", s.run.cotrans_linker)
p.addTagTarget("linker_cotrans_rc",
reverse_complement(s.run.cotrans_linker))
s.process_pair_db(pair_db)
rsid = pair_db.result_set_id_for_name(s.run.result_set_name)
pair_db.count_tags(rsid)
print pair_db.tag_counts(rsid)
def tags():
bp = "/Users/jbrink/mos/tasks/1RwIBa/tmp/datasets/cotrans/"
from spats_shape_seq.db import PairDB
pair_db = PairDB(bp + "db/pairs.db")
if True:
print "Parsing to db..."
pair_db.wipe()
pair_db.add_targets_table(bp + "cotrans_single.fa")
pair_db.parse(bp + "data/EJS_6_F_10mM_NaF_Rep1_GCCAAT_R1.fastq",
bp + "data/EJS_6_F_10mM_NaF_Rep1_GCCAAT_R2.fastq",
sample_size = 100000)
from spats_shape_seq import Spats
from spats_shape_seq.tag import TagProcessor
from spats_shape_seq.util import reverse_complement
s = Spats()
s.run._processor_class = TagProcessor
s.run.writeback_results = True
s.run.result_set_name = "tags"
s.run.num_workers = 1
s.run.cotrans = True
s.run.cotrans_linker = 'CTGACTCGGGCACCAAGGAC'
s.loadTargets(pair_db)
s.run.allow_indeterminate = True
s.run.allowed_target_errors = 2
s.run.allowed_adapter_errors = 2
p = s._processor
for target in pair_db.targets():
p.addTagTarget(target[0], target[1])
p.addTagTarget(target[0] + "_rc", reverse_complement(target[1]))
p.addTagTarget("adapter_t_rc", reverse_complement(s.run.adapter_t))
p.addTagTarget("adapter_b", s.run.adapter_b)
if s.run.cotrans:
p.addTagTarget("linker_cotrans", s.run.cotrans_linker)
p.addTagTarget("linker_cotrans_rc", reverse_complement(s.run.cotrans_linker))
s.process_pair_db(pair_db)
rsid = pair_db.result_set_id_for_name(s.run.result_set_name)
pair_db.count_tags(rsid)
print pair_db.tag_counts(rsid)
def _loadDBAndModel(self):
rsnames = self._db.result_sets()
if rsnames:
self.result_set_id = self._db.result_set_id_for_name(rsnames[0])
self._db.index_results()
self.has_tags = bool(self.result_set_id)
else:
self.result_set_id = -1
self.has_tags = False
self.has_counters = self._db.has_counters()
s = Spats()
self._db.load_run(s.run)
s.run._p_use_tag_processor = True
s.loadTargets(self._db)
if self.has_counters:
self._db.load_counters("spats", s.counters)
if self.has_tags:
p = s._processor
for t in s._targets.targets:
p.addTagTarget(t.name, t.seq)
p.addTagTarget(t.name + "_rc", reverse_complement(t.seq))
self.colors._colors[t.name.lower()] = self.colors.color("target")
p.addTagTarget("adapter_t_rc", reverse_complement(s.run.adapter_t))
p.addTagTarget("adapter_b", s.run.adapter_b)
if s.run.cotrans:
p.addTagTarget("linker_cotrans", s.run.cotrans_linker)
p.addTagTarget("linker_cotrans_rc", reverse_complement(s.run.cotrans_linker))
if s.run._p_extra_tags:
for tag, seq in s.run._p_extra_tags.iteritems():
p.addTagTarget(tag, seq)
if not self.has_counters:
p.counters.load_from_db_data(self._db.counter_data_for_results(self.result_set_id))
self._spats = s