def test_no_mrca():
seqaln = "tests/data/tiny_test_example/test.fas"
mattype = "fasta"
trfn = "tests/data/tiny_test_example/test.tre"
schema_trf = "newick"
id_to_spn = r"tests/data/tiny_test_example/test_nicespl.csv"
workdir = "tests/output/test_mrcalist_local"
configfi = "tests/data/test.config"
otu_jsonfi = "{}/otu_dict.json".format(workdir)
ingroup_mrca = None
# setup the run
if not os.path.exists("{}".format(workdir)):
os.makedirs("{}".format(workdir))
conf = ConfigObj(configfi)
ids = IdDicts(conf, workdir=workdir)
# print(ids.mrca_ott, ids.mrca_ncbi)
data_obj = pickle.load(open("tests/data/precooked/tiny_dataobj.p", 'rb'))
filteredScrape = PhyscraperScrape(data_obj, ids, ingroup_mrca)
filteredScrape.threshold = 5
assert filteredScrape.mrca_ncbi == 18794
blast_dir = "tests/data/precooked/fixed/tte_blast_files"
filteredScrape._blasted = 1
filteredScrape.read_blast_wrapper(blast_dir=blast_dir)
filteredScrape.remove_identical_seqs()
assert len(filteredScrape.new_seqs_otu_id) in [23,17] #Blurghhh, local vs remote searches get diffenrt number of seqs!
def test_add_all():
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._blasted = 1
filteredScrape.threshold = threshold
filteredScrape.read_blast_wrapper(
blast_dir="tests/data/precooked/fixed/tte_blast_files")
filteredScrape.seq_filter = [
'deleted', 'subsequence,', 'not', "removed", "deleted,"
]
filteredScrape.remove_identical_seqs()
sp_d = filteredScrape.make_sp_dict(filteredScrape.new_seqs_otu_id)
assert len(sp_d) == 5
for taxon in sp_d:
assert len(sp_d[taxon]) <= threshold
