def test_update_plus(airr_reannotated):
vdj = ddl.Dandelion(airr_reannotated)
vdj.update_plus()
assert 'mu_count' in vdj.metadata
vdj.update_plus(option='sequence')
assert 'sequence_VDJ' in vdj.metadata
vdj.update_plus(option='cdr3 lengths')
assert 'junction_aa_length_VDJ' in vdj.metadata
vdj = ddl.Dandelion(airr_reannotated)
vdj.update_plus(option='mutations')
assert 'mu_count' in vdj.metadata
vdj.update_plus(option='all')
assert 'sequence_VDJ' in vdj.metadata
vdj.update_plus(option='cdr3 lengths')
def test_travdv_filter(create_testfolder, dummy_adata_travdv):
vdj = ddl.Dandelion(
str(create_testfolder) + '/dandelion/filtered_contig_dandelion.tsv')
assert vdj.data.shape[0] == 23
assert len([i for i in vdj.data['locus'] if i == 'TRD']) == 0
vdj2, adata = ddl.pp.filter_contigs(vdj, dummy_adata_travdv)
assert vdj2.data.shape[0] == 8
def to_dandelion(adata: AnnData):
"""Export data to `Dandelion <https://github.com/zktuong/dandelion>`_ (:cite:`Stephenson2021`).
Parameters
----------
adata
annotated data matrix with :term:`IR` annotations.
Returns
-------
`Dandelion` object.
"""
try:
import dandelion as ddl
except:
raise ImportError("Please install dandelion: pip install sc-dandelion.")
airr_cells = to_airr_cells(adata)
contig_dicts = {}
for tmp_cell in airr_cells:
for i, chain in enumerate(tmp_cell.to_airr_records(), start=1):
# dandelion-specific modifications
chain.update(
{
"sequence_id": f"{tmp_cell.cell_id}_contig_{i}",
}
)
contig_dicts[chain["sequence_id"]] = chain
data = pd.DataFrame.from_dict(contig_dicts, orient="index")
return ddl.Dandelion(ddl.load_data(data))
def test_mutation(create_testfolder, airr_reannotated):
f = create_testfolder / "test.tsv"
airr_reannotated.to_csv(f, sep='\t', index=False)
ddl.pp.quantify_mutations(f)
out = pd.read_csv(f, sep='\t')
vdj = ddl.Dandelion(out)
assert not vdj.data.mu_count.empty
ddl.pp.quantify_mutations(f, frequency=True)
assert not vdj.data.mu_freq.empty
def test_manual_threshold_and_define_clones(create_testfolder):
f = create_testfolder / "test.tsv"
out = pd.read_csv(f, sep='\t')
vdj = ddl.Dandelion(out)
vdj.threshold = 0.1
ddl.tl.define_clones(vdj)
assert not vdj.data.clone_id.empty
ddl.tl.define_clones(vdj, key_added='changeo_clone')
assert not vdj.data.changeo_clone.empty
def test_quantify_mut_2(create_testfolder, processed_files, freq, colname):
f = create_testfolder / str('dandelion/' + processed_files['filtered'])
vdj = ddl.Dandelion(f)
ddl.pp.quantify_mutations(vdj, frequency=freq)
assert not vdj.data[colname].empty
if colname == 'mu_freq':
assert vdj.data[colname].dtype == float
else:
assert vdj.data[colname].dtype == int
def test_setup(create_testfolder, airr_reannotated, dummy_adata):
vdj, adata = ddl.pp.filter_contigs(airr_reannotated, dummy_adata)
assert airr_reannotated.shape[0] == 9
assert vdj.data.shape[0] == 7
assert vdj.metadata.shape[0] == 4
assert adata.n_obs == 5
vdj.data['clone_id'] = ['A', 'A', 'A', 'A', 'A', 'A', 'A']
vdj = ddl.Dandelion(vdj.data)
ddl.tl.generate_network(vdj)
ddl.tl.transfer(adata, vdj)
assert 'clone_id' in adata.obs
assert 'X_vdj' in adata.obsm
f1 = create_testfolder / "test.h5"
f2 = create_testfolder / "test.h5ad"
vdj.write_h5(f1)
adata.write_h5ad(f2)
def test_container():
os.system(
"cd /tests; python /share/dandelion_preprocess.py --meta test.csv;")
dat = pd.read_csv(
'/tests/sample_test_10x/dandelion/filtered_contig_dandelion.tsv',
sep='\t')
assert not dat['c_call'].empty
assert not dat['v_call_genotyped'].empty
assert not dat['mu_count'].empty
assert not dat['mu_freq'].empty
vdj = None
try:
vdj = ddl.Dandelion(dat)
except:
pass
assert vdj is not None
def test_generate_network(create_testfolder, resample, expected):
f = create_testfolder / "test.h5"
vdj = ddl.read_h5(f)
if resample is not None:
vdj = ddl.tl.generate_network(vdj, downsample=resample)
else:
ddl.tl.generate_network(vdj)
assert vdj.distance is not None
assert vdj.edges is None
assert vdj.n_obs == expected
assert vdj.layout is not None
assert vdj.graph is not None
vdj.data['clone_id'] = '1'
vdj = ddl.Dandelion(vdj.data)
assert vdj.data.clone_id.dtype == 'object'
ddl.tl.generate_network(vdj)
assert vdj.edges is not None
def test_update_germlines(create_testfolder, processed_files, database_paths,
fasta_10x):
f = create_testfolder / str('dandelion/' + processed_files['filtered'])
vdj = ddl.Dandelion(f)
vdj.update_germline(germline=database_paths['germline'])
assert len(vdj.germline) > 0
out_file = str(create_testfolder) + "/test_airr_reannotated.h5"
vdj.write_h5(out_file)
tmp = ddl.read_h5(out_file)
assert len(tmp.germline) > 0
vdj.update_germline(germline=database_paths['germline'],
corrected=str(create_testfolder) +
"/filtered_contig.fasta")
assert len(vdj.germline) > 0
vdj.update_germline(germline=database_paths['germline'],
corrected=fasta_10x)
assert len(vdj.germline) > 0
with pytest.raises(TypeError):
vdj.update_germline(germline=database_paths['germline'], corrected=[])
def test_setup():
file = "https://cf.10xgenomics.com/samples/cell-vdj/5.0.0/sc5p_v2_hs_B_1k_multi_5gex_b/sc5p_v2_hs_B_1k_multi_5gex_b_vdj_b_airr_rearrangement.tsv"
r = requests.get(file)
test_data = pd.read_csv(StringIO(r.text), sep="\t")
test_data["locus"] = [
"IGH" if "IGH" in i else
"IGK" if "IGK" in i else "IGL" if "IGL" in i else None
for i in test_data.v_call
]
test_data["umi_count"] = test_data["duplicate_count"]
test_data["sample_id"] = "test"
test_ddl = ddl.Dandelion(test_data)
test_ddl.write_h5("tests/test.h5", compression="bzip2")
test_ddl.write_pkl("tests/test.pkl.pbz2")
test = ddl.read_h5("tests/test.h5")
_ = ddl.read_pkl("tests/test.pkl.pbz2")
scfile = "https://cf.10xgenomics.com/samples/cell-vdj/5.0.0/sc5p_v2_hs_B_1k_multi_5gex_b/sc5p_v2_hs_B_1k_multi_5gex_b_count_filtered_feature_bc_matrix.h5"
r = requests.get(scfile)
open("tests/sctest.h5", "wb").write(r.content)
adata = sc.read_10x_h5("tests/sctest.h5")
adata.write("tests/sctest.h5ad", compression="gzip")
print(test)
print(adata)
def test_readwrite_h5(create_testfolder):
out_file1 = str(create_testfolder) + "/test_airr_reannotated.tsv"
out_file2 = str(create_testfolder) + "/test_airr_reannotated.h5"
vdj = ddl.Dandelion(out_file1)
assert not vdj.data.np1_length.empty
assert not vdj.data.np2_length.empty
assert not vdj.data.junction_length.empty
vdj.write_h5(out_file2)
vdj2 = ddl.read_h5(out_file2)
assert not vdj2.data.np1_length.empty
assert not vdj2.data.np2_length.empty
assert not vdj2.data.junction_length.empty
vdj.write_h5(out_file2, complib='blosc:lz4')
vdj2 = ddl.read_h5(out_file2)
assert not vdj2.data.np1_length.empty
assert not vdj2.data.np2_length.empty
assert not vdj2.data.junction_length.empty
vdj.write_h5(out_file2, compression='blosc:lz4')
vdj2 = ddl.read_h5(out_file2)
assert not vdj2.data.np1_length.empty
assert not vdj2.data.np2_length.empty
assert not vdj2.data.junction_length.empty
with pytest.raises(ValueError):
vdj.write_h5(out_file2, complib='blosc:lz4', compression='blosc:lz4')
def test_readwrite_pkl(create_testfolder):
out_file1 = str(create_testfolder) + "/test_airr_reannotated.tsv"
out_file2 = str(create_testfolder) + "/test_airr_reannotated.pkl"
out_file3 = str(create_testfolder) + "/test_airr_reannotated.pkl.gz"
out_file4 = str(create_testfolder) + "/test_airr_reannotated.pkl.pbz2"
vdj = ddl.Dandelion(out_file1)
assert not vdj.data.np1_length.empty
assert not vdj.data.np2_length.empty
assert not vdj.data.junction_length.empty
vdj.write_pkl(out_file2)
vdj3 = ddl.read_pkl(out_file2)
assert not vdj3.data.np1_length.empty
assert not vdj3.data.np2_length.empty
assert not vdj3.data.junction_length.empty
vdj.write_pkl(out_file3)
vdj4 = ddl.read_pkl(out_file3)
assert not vdj4.data.np1_length.empty
assert not vdj4.data.np2_length.empty
assert not vdj4.data.junction_length.empty
vdj.write_pkl(out_file4)
vdj5 = ddl.read_pkl(out_file4)
assert not vdj5.data.np1_length.empty
assert not vdj5.data.np2_length.empty
assert not vdj5.data.junction_length.empty
def test_update_germlines(create_testfolder, processed_files, database_paths):
f = create_testfolder / str('dandelion/' + processed_files['filtered'])
vdj = ddl.Dandelion(f)
vdj.update_germlines(database_paths['germline'])
assert len(vdj.germline) > 0
def test_loadtravdv2(airr_travdv):
vdj = ddl.Dandelion(airr_travdv)
assert vdj.data.shape[0] == 6
assert all([i == 'TRD' for i in vdj.data['locus']])
def test_create_germlines(create_testfolder, database_paths):
f = create_testfolder / "test.tsv"
out = pd.read_csv(f, sep='\t')
vdj = ddl.Dandelion(out)
ddl.pp.create_germlines(vdj, germline=database_paths['germline'])
assert not vdj.data.germline_alignment_d_mask.empty
def test_loadtravdv_reannotated(create_testfolder):
vdj = ddl.Dandelion(
str(create_testfolder) + '/dandelion/filtered_contig_dandelion.tsv')
assert vdj.data.shape[0] == 23
assert len([i for i in vdj.data['locus'] if i == 'TRD']) == 0
def test_update_germlines_fail(create_testfolder, processed_files):
f = create_testfolder / str('dandelion/' + processed_files['filtered'])
vdj = ddl.Dandelion(f)
with pytest.raises(KeyError):
vdj.update_germline()
