def test_filter_length():
workdir = "tests/output/test_selectbylength"
absworkdir = os.path.abspath(workdir)
conf = ConfigObj("tests/data/test.config", interactive=False)
threshold = 2
selectby = "length"
downtorank = "species"
add_unpubl_seq = None
blacklist = None
id_to_spn_addseq_json = None
ingroup_mrca = None
shared_blast_folder = None
data_obj = pickle.load(open("tests/data/precooked/tiny_dataobj.p", 'rb'))
data_obj.workdir = absworkdir
ids = IdDicts(conf, workdir=data_obj.workdir)
ids.acc_ncbi_dict = pickle.load(
open("tests/data/precooked/tiny_acc_map.p", "rb"))
# Now combine the data, the ids, and the configuration into a single physcraper scrape object
filteredScrape = PhyscraperScrape(data_obj, ids)
filteredScrape.blacklist = blacklist
sys.stdout.write("BLASTing input sequences\n")
if shared_blast_folder:
filteredScrape.blast_subdir = shared_blast_folder
else:
shared_blast_folder = None
# filteredScrape.run_blast_wrapper()
filteredScrape.read_blast_wrapper(
blast_dir="tests/data/precooked/fixed/tte_blast_files")
filteredScrape.remove_identical_seqs()
filteredScrape.dump()
sys.stdout.write("Filter the sequences\n")
length_unfiltered = len(filteredScrape.new_seqs_otu_id)
# if threshold is not None:
# filteredScrape.filter_seqs()
length_filtered = len(filteredScrape.new_seqs)
def test_add_local():
data_obj = pickle.load(open("tests/data/precooked/tiny_dataobj.p", 'rb'))
data_obj.workdir = absworkdir
ids = IdDicts(conf, workdir=data_obj.workdir)
ids.acc_ncbi_dict = pickle.load(open("tests/data/precooked/tiny_acc_map.p", "rb"))
if not os.path.exists("{}".format(workdir)):
os.makedirs("{}".format(workdir))
if os.path.exists(otu_jsonfi_local):
otu_json_local = json.load(open(otu_jsonfi_local))
else:
otu_json_local = OtuJsonDict(id_to_spn_addseq, ids)
json.dump(otu_json_local, open(otu_jsonfi_local, "w"))
# Now combine the data, the ids, and the configuration into a single physcraper scrape object
filteredScrape = PhyscraperScrape(data_obj, ids)
filteredScrape.blacklist = blacklist
if add_local_seq is not None:
filteredScrape.unpublished = True
if filteredScrape.unpublished is True: # use unpublished data
# filteredScrape.unpublished = True
filteredScrape.data.unpubl_otu_json = otu_json_local
filteredScrape.write_unpubl_blastdb(add_local_seq)
# filteredScrape.make_otu_dict_entry_unpubl()
filteredScrape.run_blast_wrapper()
filteredScrape.read_blast_wrapper()
filteredScrape.remove_identical_seqs()
test = False
for key in filteredScrape.data.otu_dict.keys():
if '^ncbi:title' in filteredScrape.data.otu_dict[key].keys():
if filteredScrape.data.otu_dict[key]['^ncbi:title'] == "unpublished":
test = True
break
assert test == True
def test_blacklist():
workdir = "tests/output/test_blacklist"
configfi = "tests/data/test.config"
# make one run without blacklist
blacklist = None
noblack = os.path.join(workdir, "noblacklist")
absworkdir = os.path.abspath(noblack)
if not os.path.exists(os.path.join(absworkdir, "current_blast_run/")):
os.makedirs(os.path.join(absworkdir, "current_blast_run/"))
conf = ConfigObj(configfi, interactive=False)
data_obj = pickle.load(open("tests/data/precooked/tiny_dataobj.p", 'rb'))
data_obj.workdir = absworkdir
ids = IdDicts(conf, workdir=data_obj.workdir)
ids.acc_ncbi_dict = pickle.load(open("tests/data/precooked/tiny_acc_map.p", "rb"))
noblackScrape = PhyscraperScrape(data_obj, ids)
noblackScrape._blasted = 1
src = "tests/data/precooked/fixed/tte_blast_files"
src_files = os.listdir(src)
for file_name in src_files:
dest = os.path.join(absworkdir, "current_blast_run/")
# print(dest)
full_file_name = os.path.join(src, file_name)
if (os.path.isfile(full_file_name)):
shutil.copy(full_file_name, dest)
noblackScrape.read_blast_wrapper()
noblackScrape.remove_identical_seqs()
new_test_generate_streamed_aln(noblackScrape)
# one run with blacklist
blacklist = ['JX895340.1']
absworkdir = os.path.abspath(workdir)
conf = ConfigObj(configfi, interactive=False)
data_obj = pickle.load(open("tests/data/precooked/tiny_dataobj.p", 'rb'))
data_obj.workdir = absworkdir
ids = IdDicts(conf, workdir=data_obj.workdir)
ids.acc_ncbi_dict = pickle.load(open("tests/data/precooked/tiny_acc_map.p", "rb"))
filteredScrape = PhyscraperScrape(data_obj, ids)
filteredScrape.blacklist = blacklist
filteredScrape._blasted = 1
if not os.path.exists(os.path.join(absworkdir, "current_blast_run/")):
os.makedirs(os.path.join(absworkdir, "current_blast_run/"))
src = "tests/data/precooked/fixed/tte_blast_files"
src_files = os.listdir(src)
for file_name in src_files:
dest = os.path.join(absworkdir, "current_blast_run/")
full_file_name = os.path.join(src, file_name)
if (os.path.isfile(full_file_name)):
shutil.copy(full_file_name, dest)
# filteredScrape.acc_list_mrca = pickle.load(open("tests/data/precooked/acc_list_mrca.p", 'rb'))
filteredScrape.read_blast_wrapper()
filteredScrape.remove_identical_seqs()
new_test_generate_streamed_aln(filteredScrape)
print("RUN TESTS!")
gi_l = []
gi_l_2 = []
for tax in filteredScrape.data.tre.taxon_namespace:
gi_id = filteredScrape.data.otu_dict[tax.label].get("^ncbi:accession")
gi_l.append(gi_id)
print(gi_l)
for tax in noblackScrape.data.tre.taxon_namespace:
# print(filteredScrape.data.otu_dict[tax.label])
gi_id = noblackScrape.data.otu_dict[tax.label].get("^ncbi:accession")
gi_l_2.append(gi_id)
print(gi_l_2)
for item in blacklist:
assert item not in gi_l
