def test_read_10x_csv():
anndata = read_10x_vdj(TESTDATA / "10x/filtered_contig_annotations.csv")
obs = anndata.obs
assert obs.shape[0] == 5
cell1 = obs.iloc[1, :]
cell2 = obs.iloc[3, :]
cell3 = obs.iloc[4, :]
assert cell1.name == "AAACCTGAGTACGCCC-1"
assert cell1["IR_VDJ_1_junction_aa"] == "CASSLGPSTDTQYF"
assert cell1[
"IR_VDJ_1_junction"] == "TGTGCCAGCAGCTTGGGACCTAGCACAGATACGCAGTATTTT"
assert cell1["IR_VDJ_1_duplicate_count"] == 55
assert cell1["IR_VDJ_1_consensus_count"] == 18021
assert cell1["IR_VDJ_1_v_call"] == "TRBV7-2"
assert cell1["IR_VDJ_1_d_call"] == "TRBD2"
assert cell1["IR_VDJ_1_j_call"] == "TRBJ2-3"
assert cell1["IR_VDJ_1_c_call"] == "TRBC2"
assert _is_false(cell1["multi_chain"])
assert cell1["IR_VJ_1_locus"] == "TRA"
assert cell1["IR_VDJ_1_locus"] == "TRB"
assert cell2.name == "AAACCTGGTCCGTTAA-1"
assert cell2["IR_VJ_1_junction_aa"] == "CALNTGGFKTIF"
assert cell2["IR_VJ_2_junction_aa"] == "CAVILDARLMF"
assert cell2["IR_VJ_1_duplicate_count"] == 5
assert cell2["IR_VJ_2_duplicate_count"] == 5
assert cell2["IR_VJ_1_locus"] == "TRA"
assert cell2["IR_VDJ_1_locus"] == "TRB"
assert cell2["IR_VJ_2_locus"] == "TRA"
assert _is_na(cell2["IR_VDJ_2_junction_aa"])
assert cell3.name == "AAACTTGGTCCGTTAA-1"
assert cell3["IR_VJ_1_locus"] == "IGK"
assert cell3["IR_VDJ_1_locus"] == "IGH"
def test_read_10x():
anndata = read_10x_vdj("tests/data/10x/all_contig_annotations.json")
obs = anndata.obs
# this has `is_cell=false` and should be filtered out
assert "AAACCTGAGACCTTTG-1" not in anndata.obs_names
assert obs.shape[0] == 2
cell1 = obs.iloc[0, :]
cell2 = obs.iloc[1, :]
assert cell1.name == "AAACCTGAGACCTTTG-2"
assert cell1["TRB_1_cdr3"] == "CASSPPSQGLSTGELFF"
assert (
cell1["TRB_1_cdr3_nt"] == "TGTGCCAGCTCACCACCGAGCCAGGGCCTTTCTACCGGGGAGCTGTTTTTT"
)
assert cell1["TRB_1_junction_ins"] == 4 + 7
assert cell1["TRB_1_expr"] == 1
assert cell1["TRB_1_v_gene"] == "TRBV18"
assert cell1["TRB_1_d_gene"] == "TRBD1"
assert cell1["TRB_1_j_gene"] == "TRBJ2-2"
assert cell1["TRB_1_c_gene"] == "TRBC2"
assert _is_false(cell1["multi_chain"])
assert np.all(_is_na(cell1[["TRA_1_cdr3", "TRB_2_cdr3", "TRA_1_junction_ins"]]))
assert cell2.name == "AAACCTGAGTACGCCC-1"
assert cell2["TRA_1_cdr3"] == "CAMRVGGSQGNLIF"
assert cell2["TRA_2_cdr3"] == "CATDAKDSNYQLIW"
assert cell2["TRA_1_expr"] == 9
assert cell2["TRA_2_expr"] == 4
assert np.all(_is_na(cell2[["TRB_1_cdr3", "TRB_2_cdr3"]]))
assert cell2["TRA_1_junction_ins"] == 4
assert cell2["TRA_2_junction_ins"] == 4
def test_read_10x_example():
"""Test that a full 10x CSV table can be imported without errors.
Test-dataset from https://support.10xgenomics.com/single-cell-vdj/datasets/3.1.0/vdj_nextgem_hs_pbmc3
under CC-BY-4.0
"""
anndata = read_10x_vdj(
"tests/data/10x/vdj_nextgem_hs_pbmc3_t_filtered_contig_annotations.csv.gz"
)
def test_read_10x_json_cr6():
"""Test additional cols from CR6 outputs: fwr{1,2,3,4}{,_nt} and cdr{1,2}{,_nt}"""
anndata = read_10x_vdj(
TESTDATA /
"10x/10k_BMMNC_5pv2_nextgem_Multiplex_vdj_t_all_contig_annotations_small.json",
include_fields=None,
)
obs = anndata.obs
assert obs.shape[0] == 2
cell1 = obs.iloc[0, :]
assert cell1.name == "AAACCTGCACAGGTTT-1"
assert cell1["IR_VDJ_1_fwr1_aa"] == "KAGVTQTPRYLIKTRGQQVTLSCSPI"
assert (
cell1["IR_VDJ_1_fwr1"] ==
"AAGGCTGGAGTCACTCAAACTCCAAGATATCTGATCAAAACGAGAGGACAGCAAGTGACACTGAGCTGCTCCCCTATC"
)
assert cell1["IR_VDJ_1_cdr1_aa"] == "SGHRS"
assert cell1["IR_VDJ_1_cdr1"] == "TCTGGGCATAGGAGT"
assert cell1["IR_VDJ_1_fwr2_aa"] == "VSWYQQTPGQGLQFLFE"
assert (cell1["IR_VDJ_1_fwr2"] ==
"GTATCCTGGTACCAACAGACCCCAGGACAGGGCCTTCAGTTCCTCTTTGAA")
assert cell1["IR_VDJ_1_cdr2_aa"] == "YFSETQ"
assert cell1["IR_VDJ_1_cdr2"] == "TACTTCAGTGAGACACAG"
assert cell1["IR_VDJ_1_fwr3_aa"] == "RNKGNFPGRFSGRQFSNSRSEMNVSTLELGDSALYL"
assert (
cell1["IR_VDJ_1_fwr3"] ==
"AGAAACAAAGGAAACTTCCCTGGTCGATTCTCAGGGCGCCAGTTCTCTAACTCTCGCTCTGAGATGAATGTGAGCACCTTGGAGCTGGGGGACTCGGCCCTTTATCTT"
)
assert cell1["IR_VDJ_1_cdr3_aa"] == "ASSWMDRGEAF"
assert cell1["IR_VDJ_1_cdr3"] == "GCCAGCAGCTGGATGGATAGGGGTGAAGCTTTC"
assert cell1["IR_VDJ_1_fwr4_aa"] == "GQGTRLTVV"
assert cell1["IR_VDJ_1_fwr4"] == "GGACAAGGCACCAGACTCACAGTTGTAG"
assert cell1["IR_VJ_1_fwr1_aa"] == "AQTVTQSQPEMSVQEAETVTLSCTYD"
assert (
cell1["IR_VJ_1_fwr1"] ==
"GCTCAGACAGTCACTCAGTCTCAACCAGAGATGTCTGTGCAGGAGGCAGAGACCGTGACCCTGAGCTGCACATATGAC"
)
assert cell1["IR_VJ_1_cdr1_aa"] == "TSESDYY"
assert cell1["IR_VJ_1_cdr1"] == "ACCAGTGAGAGTGATTATTAT"
assert cell1["IR_VJ_1_fwr2_aa"] == "LFWYKQPPSRQMILVIR"
assert (cell1["IR_VJ_1_fwr2"] ==
"TTATTCTGGTACAAGCAGCCTCCCAGCAGGCAGATGATTCTCGTTATTCGC")
assert cell1["IR_VJ_1_cdr2_aa"] == "QEAYKQQN"
assert cell1["IR_VJ_1_cdr2"] == "CAAGAAGCTTATAAGCAACAGAAT"
assert cell1["IR_VJ_1_fwr3_aa"] == "ATENRFSVNFQKAAKSFSLKISDSQLGDAAMYF"
assert (
cell1["IR_VJ_1_fwr3"] ==
"GCAACAGAGAATCGTTTCTCTGTGAACTTCCAGAAAGCAGCCAAATCCTTCAGTCTCAAGATCTCAGACTCACAGCTGGGGGATGCCGCGATGTATTTC"
)
assert cell1["IR_VJ_1_cdr3_aa"] == "ALYKVTGNQFY"
assert cell1["IR_VJ_1_cdr3"] == "GCTCTTTATAAGGTCACCGGTAACCAGTTCTAT"
assert cell1["IR_VJ_1_fwr4_aa"] == "GTGTSLTVIP"
assert cell1["IR_VJ_1_fwr4"] == "GGGACAGGGACAAGTTTGACGGTCATTCCAA"
def _read_anndata_from_10x_sample(path):
"""Read full 10x CSV table and convert it to IR objects, ready
to be used for roundtrip conversions.
Test-dataset from https://support.10xgenomics.com/single-cell-vdj/datasets/3.1.0/vdj_nextgem_hs_pbmc3
and https://support.10xgenomics.com/single-cell-vdj/datasets/4.0.0/sc5p_v2_hs_melanoma_10k
under CC-BY-4.0.
Pytest only caches one fixture at a time, i.e. it doesn't work with parametrized
fixtures. Therefore, we use the lru_cache instead.
"""
print(f"Reading 10x file: {path}")
anndata = read_10x_vdj(path, include_fields=None)
return anndata
def test_read_10x():
anndata = read_10x_vdj(TESTDATA / "10x/all_contig_annotations.json",
include_fields=None)
obs = anndata.obs
# this has `is_cell=false` and should be filtered out
assert "AAACCTGAGACCTTTG-1" not in anndata.obs_names
assert obs.shape[0] == 3
cell1 = obs.iloc[0, :]
cell2 = obs.iloc[1, :]
cell3 = obs.iloc[2, :]
assert cell1.name == "AAACCTGAGACCTTTG-2"
assert cell1["IR_VDJ_1_junction_aa"] == "CASSPPSQGLSTGELFF"
assert (cell1["IR_VDJ_1_junction"] ==
"TGTGCCAGCTCACCACCGAGCCAGGGCCTTTCTACCGGGGAGCTGTTTTTT")
assert cell1["IR_VDJ_1_np1_length"] == 4
assert cell1["IR_VDJ_1_np2_length"] == 7
assert cell1["IR_VDJ_1_duplicate_count"] == 1
assert cell1["IR_VDJ_1_consensus_count"] == 494
assert cell1["IR_VDJ_1_v_call"] == "TRBV18"
assert cell1["IR_VDJ_1_d_call"] == "TRBD1"
assert cell1["IR_VDJ_1_j_call"] == "TRBJ2-2"
assert cell1["IR_VDJ_1_c_call"] == "TRBC2"
assert _is_false(cell1["multi_chain"])
assert np.all(
_is_na(cell1[[
"IR_VJ_1_junction_aa", "IR_VDJ_2_junction_aa", "IR_VJ_1_np1_length"
]]))
assert cell2.name == "AAACCTGAGTACGCCC-1"
assert cell2["IR_VJ_1_junction_aa"] == "CAMRVGGSQGNLIF"
assert cell2["IR_VJ_2_junction_aa"] == "CATDAKDSNYQLIW"
assert cell2["IR_VJ_1_duplicate_count"] == 9
assert cell2["IR_VJ_2_duplicate_count"] == 4
assert np.all(
_is_na(cell2[["IR_VDJ_1_junction_aa", "IR_VDJ_2_junction_aa"]]))
assert cell2["IR_VJ_1_np1_length"] == 4
assert _is_na(cell2["IR_VJ_1_np2_length"])
assert cell2["IR_VJ_2_np1_length"] == 4
assert _is_na(cell2["IR_VJ_2_np2_length"])
assert cell3.name == "CAGGTGCTCGTGGTCG-1"
assert cell3["IR_VJ_1_locus"] == "IGK"
assert _is_na(cell3["IR_VJ_2_locus"]) # non-productive
assert cell3["IR_VDJ_1_locus"] == "IGH"
assert _is_na(cell3["IR_VDJ_2_locus"]) # non-productive
def test_read_and_convert_10x_example(path):
"""Test that a full 10x CSV table can be imported without errors.
Additionally test that the round-trip conversion using `to_ir_objs` and
`from_ir_objs` is the identity. Doing this here to avoid loading the data twice
since this is already one of the longer-running tests.
Test-dataset from https://support.10xgenomics.com/single-cell-vdj/datasets/3.1.0/vdj_nextgem_hs_pbmc3
and https://support.10xgenomics.com/single-cell-vdj/datasets/4.0.0/sc5p_v2_hs_melanoma_10k
under CC-BY-4.0
"""
anndata = read_10x_vdj(path)
assert anndata.shape[0] > 0
# Test that round-trip conversion succeeds
ir_objs = to_ir_objs(anndata)
anndata2 = from_ir_objs(ir_objs)
pdt.assert_frame_equal(anndata.obs, anndata2.obs)
def test_read_10x_csv():
anndata = read_10x_vdj("tests/data/10x/filtered_contig_annotations.csv")
obs = anndata.obs
assert obs.shape[0] == 4
cell1 = obs.iloc[1, :]
cell2 = obs.iloc[3, :]
assert cell1.name == "AAACCTGAGTACGCCC-1"
assert cell1["TRB_1_cdr3"] == "CASSLGPSTDTQYF"
assert cell1["TRB_1_cdr3_nt"] == "TGTGCCAGCAGCTTGGGACCTAGCACAGATACGCAGTATTTT"
assert _is_na(cell1["TRB_1_junction_ins"])
assert cell1["TRB_1_expr"] == 55
assert cell1["TRB_1_v_gene"] == "TRBV7-2"
assert cell1["TRB_1_d_gene"] == "TRBD2"
assert cell1["TRB_1_j_gene"] == "TRBJ2-3"
assert cell1["TRB_1_c_gene"] == "TRBC2"
assert _is_false(cell1["multi_chain"])
assert cell2.name == "AAACCTGGTCCGTTAA-1"
assert cell2["TRA_1_cdr3"] == "CALNTGGFKTIF"
assert cell2["TRA_2_cdr3"] == "CAVILDARLMF"
assert cell2["TRA_1_expr"] == 5
assert cell2["TRA_2_expr"] == 5
assert _is_na(cell2["TRB_2_cdr3"])
