loqusdb 2.7.2__py3-none-any.whl → 2.7.8__py3-none-any.whl

tests/vcf_tools/test_check_vcf.py

@@ -0,0 +1,64 @@
+import pytest
+from loqusdb.exceptions import VcfError
+from loqusdb.utils.vcf import check_vcf
+
+
+def test_check_vcf_correct(vcf_path):
+    ## GIVEN a vcf file and a counter that checks the number of variants
+    true_nr = 0
+
+    with open(vcf_path, "r") as f:
+        for line in f:
+            if not line.startswith("#"):
+                true_nr += 1
+
+    ## WHEN collecting the VCF info
+    vcf_info = check_vcf(vcf_path)
+
+    ## THEN assert that the number of variants collected is correct
+    assert vcf_info["nr_variants"] == true_nr
+    ## THEN assert that the variant type is correct
+    assert vcf_info["variant_type"] == "snv"
+
+
+def test_check_vcf_double_variant(double_vcf_path):
+    ## GIVEN a variant file where a variant is duplicated
+    ## WHEN checking the vcf
+    ## THEN assert that the function raises a VcfError
+    with pytest.raises(VcfError):
+        check_vcf(double_vcf_path)
+
+
+def test_check_vcf_unsorted(unsorted_vcf_path):
+    ## GIVEN a vcf file with unsorted variants
+    ## WHEN checking the vcf
+    ## THEN assert that the function raises a VcfError
+    with pytest.raises(VcfError):
+        check_vcf(unsorted_vcf_path)
+
+
+def test_check_sv_vcf(sv_vcf_path):
+    ## GIVEN a vcf file and a counter that checks the number of variants
+    true_nr = 0
+
+    with open(sv_vcf_path, "r") as f:
+        for line in f:
+            if not line.startswith("#"):
+                true_nr += 1
+
+    ## WHEN collecting the VCF info
+    vcf_info = check_vcf(sv_vcf_path, "sv")
+
+    ## THEN assert that the number of variants collected is correct
+    assert vcf_info["nr_variants"] == true_nr
+    ## THEN assert that the variant type is correct
+    assert vcf_info["variant_type"] == "sv"
+
+
+def test_check_vcf_wrong_type(sv_vcf_path):
+    ## GIVEN a sv vcf file
+
+    ## WHEN collecting the VCF info with wrong variant type
+    ## THEN assert that a VcfError is raised
+    with pytest.raises(VcfError):
+        vcf_info = check_vcf(sv_vcf_path, "snv")

tests/vcf_tools/test_format_sv_variant.py

@@ -0,0 +1,102 @@
+from loqusdb.build_models.variant import build_variant
+
+
+def test_format_indel(del_variant, case_obj):
+    ## GIVEN a SV deletion
+    variant = del_variant
+    case_id = case_obj["case_id"]
+    ## WHEN parsing the variant
+    formated_variant = build_variant(variant=variant, case_obj=case_obj, case_id=case_id)
+    expected_id = "_".join([variant.CHROM, str(variant.POS), variant.REF, variant.ALT[0]])
+
+    ## THEN assert the sv is parsed correct
+    assert formated_variant
+    assert formated_variant["variant_id"] == expected_id
+    assert formated_variant["chrom"] == variant.CHROM
+    assert formated_variant["end_chrom"] == variant.CHROM
+    assert formated_variant["pos"] == variant.POS
+    assert formated_variant["end"] == variant.INFO["END"]
+    assert formated_variant["sv_len"] == abs(variant.INFO["SVLEN"])
+
+    assert formated_variant["ref"] == variant.REF
+    assert formated_variant["alt"] == variant.ALT[0]
+    assert formated_variant["sv_type"] == "DEL"
+    assert formated_variant["case_id"] == case_id
+    assert formated_variant["homozygote"] == 0
+    assert formated_variant["hemizygote"] == 0
+
+
+def test_format_small_ins(small_insert_variant, case_obj):
+    ## GIVEN a small insertion (This means that the insertion is included in ALT field)
+    variant = small_insert_variant
+    case_id = case_obj["case_id"]
+    ## WHEN parsing the variant
+    formated_variant = build_variant(variant=variant, case_obj=case_obj, case_id=case_id)
+
+    ## THEN assert the sv is parsed correct
+    assert formated_variant["chrom"] == variant.CHROM
+    assert formated_variant["end_chrom"] == variant.CHROM
+    assert formated_variant["pos"] == variant.POS
+    assert formated_variant["end"] == variant.POS + abs(variant.INFO["SVLEN"])
+    assert formated_variant["sv_len"] == abs(variant.INFO["SVLEN"])
+
+    assert formated_variant["ref"] == variant.REF
+    assert formated_variant["alt"] == variant.ALT[0]
+    assert formated_variant["sv_type"] == "INS"
+
+
+def test_format_insertion(insertion_variant, case_obj):
+    ## GIVEN a small insertion (This means that the insertion is included in ALT field)
+    variant = insertion_variant
+    case_id = case_obj["case_id"]
+    ## WHEN parsing the variant
+    formated_variant = build_variant(variant=variant, case_obj=case_obj, case_id=case_id)
+
+    ## THEN assert the sv is parsed correct
+    assert formated_variant["chrom"] == variant.CHROM
+    assert formated_variant["end_chrom"] == variant.CHROM
+    assert formated_variant["pos"] == variant.POS
+    assert formated_variant["end"] == variant.INFO["END"]
+    assert formated_variant["sv_len"] == 0
+
+    assert formated_variant["ref"] == variant.REF
+    assert formated_variant["alt"] == variant.ALT[0]
+    assert formated_variant["sv_type"] == "INS"
+
+
+def test_format_dup_tandem(duptandem_variant, case_obj):
+    ## GIVEN a small insertion (This means that the insertion is included in ALT field)
+    variant = duptandem_variant
+    case_id = case_obj["case_id"]
+    ## WHEN parsing the variant
+    formated_variant = build_variant(variant=variant, case_obj=case_obj, case_id=case_id)
+
+    ## THEN assert the sv is parsed correct
+    assert formated_variant["chrom"] == variant.CHROM
+    assert formated_variant["end_chrom"] == variant.CHROM
+    assert formated_variant["pos"] == variant.POS
+    assert formated_variant["end"] == variant.INFO["END"]
+    assert formated_variant["sv_len"] == abs(variant.INFO["SVLEN"])
+
+    assert formated_variant["ref"] == variant.REF
+    assert formated_variant["alt"] == variant.ALT[0]
+    assert formated_variant["sv_type"] == "DUP"
+
+
+def test_format_translocation(translocation_variant, case_obj):
+    ## GIVEN a small insertion (This means that the insertion is included in ALT field)
+    variant = translocation_variant
+    case_id = case_obj["case_id"]
+    ## WHEN parsing the variant
+    formated_variant = build_variant(variant=variant, case_obj=case_obj, case_id=case_id)
+
+    ## THEN assert the sv is parsed correct
+    assert formated_variant["chrom"] == variant.CHROM
+    assert formated_variant["end_chrom"] == "11"
+    assert formated_variant["pos"] == variant.POS
+    assert formated_variant["end"] == 119123896
+    assert formated_variant["sv_len"] == float("inf")
+
+    assert formated_variant["ref"] == variant.REF
+    assert formated_variant["alt"] == variant.ALT[0]
+    assert formated_variant["sv_type"] == "BND"

tests/vcf_tools/test_format_variant.py

@@ -0,0 +1,113 @@
+from loqusdb.build_models.variant import build_variant, GENOTYPE_MAP
+
+
+def test_format_variant(het_variant, case_obj):
+    ## GIVEN a parsed variant
+    variant = het_variant
+    case_id = case_obj["case_id"]
+    ## WHEN parsing the variant
+    formated_variant = build_variant(variant=variant, case_obj=case_obj, case_id=case_id)
+
+    expected_id = "_".join([variant.CHROM, str(variant.POS), variant.REF, variant.ALT[0]])
+
+    ## THEN assert it was built in a correct way
+    assert formated_variant
+    assert formated_variant["variant_id"] == expected_id
+    assert formated_variant["chrom"] == variant.CHROM
+    assert formated_variant["pos"] == variant.POS
+    assert formated_variant["ref"] == variant.REF
+    assert formated_variant["alt"] == variant.ALT[0]
+    assert formated_variant["case_id"] == case_id
+    assert formated_variant["homozygote"] == 0
+
+
+def test_format_variant_no_qual(variant_no_gq, case_obj):
+    ## GIVEN a variant without GQ
+    variant = variant_no_gq
+    ## And that has a missing QUAL value
+    variant.QUAL = None
+    case_id = case_obj["case_id"]
+    ## WHEN parsing the variant using a QUAL threshold
+    formated_variant = build_variant(
+        variant=variant, case_obj=case_obj, case_id=case_id, gq_qual=True, gq_threshold=20
+    )
+    ## THEN assert that None is returned since requirements are not fulfilled
+    assert formated_variant is None
+
+
+def test_format_variant_no_gq(variant_no_gq, case_obj):
+    ## GIVEN a variant without GQ
+    variant = variant_no_gq
+    case_id = case_obj["case_id"]
+    ## WHEN parsing the variant using a GQ threshold
+    formated_variant = build_variant(
+        variant=variant, case_obj=case_obj, case_id=case_id, gq_threshold=20
+    )
+    ## THEN assert that None is returned since requirements are not fulfilled
+    assert formated_variant is None
+
+
+def test_format_variant_chr_prefix(variant_chr, case_obj):
+    ## GIVEN a variant with 'chr' prefix in chromosome name
+    variant = variant_chr
+    assert variant.CHROM.startswith("chr")
+    case_id = case_obj["case_id"]
+    ## WHEN parsing the variant using a GQ threshold
+    formated_variant = build_variant(
+        variant=variant, case_obj=case_obj, case_id=case_id, gq_threshold=20
+    )
+    ## THEN assert that the 'chr' part has been stripped away
+    assert formated_variant["chrom"] == variant.CHROM[3:]
+
+
+def test_format_variant_no_family_id(het_variant, case_obj):
+    ## GIVEN a parsed variant
+    variant = het_variant
+    case_id = case_obj["case_id"]
+    ## WHEN parsing the variant telling that 'case_id' is None
+    formated_variant = build_variant(variant=variant, case_obj=case_obj, case_id=None)
+    ## THEN assert that case_id was not added
+    assert formated_variant.get("case_id") == None
+    assert formated_variant["homozygote"] == 0
+    assert formated_variant["hemizygote"] == 0
+
+
+def test_format_homozygote_variant(hom_variant, case_obj):
+    ## GIVEN a parsed hom variant
+    variant = hom_variant
+    case_id = case_obj["case_id"]
+
+    ## WHEN parsing the variant
+    formated_variant = build_variant(variant=variant, case_obj=case_obj, case_id=case_id)
+
+    ## THEN assert that the variant has hom count
+    assert formated_variant["homozygote"] == 1
+    assert formated_variant["hemizygote"] == 0
+
+
+def test_format_hemizygote_variant(hem_variant, case_obj):
+    ## GIVEN a parsed hemizygous variant
+    variant = hem_variant
+    case_id = case_obj["case_id"]
+
+    ## WHEN parsing the variant
+    formated_variant = build_variant(variant=variant, case_obj=case_obj, case_id=case_id)
+
+    ## THEN assert that hemizygote count is 1
+    assert formated_variant["homozygote"] == 0
+    assert formated_variant["hemizygote"] == 1
+
+
+def test_format_variant_no_call(variant_no_call, case_obj):
+    ## GIVEN a parsed variant with no call in all individuals
+    variant = variant_no_call
+    case_id = case_obj["case_id"]
+
+    for call in variant.gt_types:
+        assert GENOTYPE_MAP[call] in ["no_call", "hom_ref"]
+
+    ## WHEN parsing the variant
+    formated_variant = build_variant(variant=variant, case_obj=case_obj, case_id=case_id)
+
+    ## THEN assert that the result is None
+    assert formated_variant is None

tests/vcf_tools/test_vcf.py

@@ -0,0 +1,63 @@
+import pytest
+from cyvcf2 import VCF
+from loqusdb.utils.vcf import get_file_handle, check_vcf
+
+
+def test_get_file_handle(vcf_path):
+    ## GIVEN the path to a vcf
+
+    ## WHEN geting the file handle
+    vcf = get_file_handle(vcf_path)
+
+    ## THEN assert that a VCF object is returned
+    assert type(vcf) is VCF
+
+
+def test_get_zipped_file_handle(zipped_vcf_path):
+    ## GIVEN the path to a zipped vcf
+
+    ## WHEN geting the file handle
+    vcf = get_file_handle(zipped_vcf_path)
+
+    ## THEN assert that a VCF object is returned
+    assert type(vcf) is VCF
+
+
+def test_get_vcf_non_vcf(ped_path):
+    ## GIVEN the path to a non vcf
+
+    ## WHEN geting the file handle
+
+    ## THEN assert that a IOError is raised
+    with pytest.raises(IOError):
+        vcf = get_file_handle(ped_path)
+
+
+def test_get_vcf_non_existing():
+    ## GIVEN the path to a non existing file
+
+    ## WHEN geting the file handle
+
+    ## THEN assert that a IOError is raised
+    with pytest.raises(IOError):
+        vcf = get_file_handle("hello")
+
+
+def test_check_vcf(vcf_path):
+    ## GIVEN the path to a vcf
+    nr_variants = 0
+    vcf = VCF(vcf_path)
+    inds = vcf.samples
+    for var in vcf:
+        nr_variants += 1
+    ## WHEN checking the vcf
+    vcf_info = check_vcf(vcf_path)
+
+    ## THEN assert that the number of variants is correct
+    assert vcf_info["nr_variants"] == nr_variants
+
+    ## THEN assert that the individuals are returned
+    assert vcf_info["individuals"] == inds
+
+    ## THEN assert that the variant type is correct
+    assert vcf_info["variant_type"] == "snv"

loqusdb-2.7.2.dist-info/RECORD

@@ -1,54 +0,0 @@
-loqusdb/__init__.py,sha256=mUCPADpIiji9CijrvermJNjJDgTFv8TlapV5qVyszHs,1658
-loqusdb/__main__.py,sha256=8FGKySAGaWSzAYMj6HRsxeyiME3V01Idt7HrmN7pSYY,397
-loqusdb/log.py,sha256=CDcrCjzs9ef-d5Wg8Q_41bCOZRM5j8PyP06kNcynTj0,1691
-loqusdb/build_models/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
-loqusdb/build_models/case.py,sha256=P3sfQkI_fH8u5iqecYWhV866lcHz4upWkepaea5MMIw,4255
-loqusdb/build_models/profile_variant.py,sha256=TbSxfVjESstS_FgbkOW4NQwMQVeTyhn9oc9yPZmDhzI,1021
-loqusdb/build_models/variant.py,sha256=JE3o_htRQjpwwUPxmP5HJ_0Ax8BwKp610uBIo3HTaVA,6671
-loqusdb/commands/__init__.py,sha256=BXAN3UADgqPrkGczzjlLO9GyyQ96dnLnP7n92JlYHgo,603
-loqusdb/commands/annotate.py,sha256=748kImopE5WbaO1nuv3WUgIqezWFSsi7SBeWhOz26-s,1384
-loqusdb/commands/cli.py,sha256=wJD5S1BoCxtRTAobd1QtmQpzlngJg-mt1nsyD92fDD4,3176
-loqusdb/commands/delete.py,sha256=R6ysHKSMw1mmL4ZbktoUIKzdzDLQ3314YPYhIy1myic,1979
-loqusdb/commands/export.py,sha256=0V3S3QU9LKlR13w3KCGfqFliTYiDRCgNwusA27AEvmE,3254
-loqusdb/commands/identity.py,sha256=hzbnvniKgSNEwSeYHsxdNvVlqu_vXeOSLlNFnDXTQjA,779
-loqusdb/commands/load.py,sha256=cVDdY7meBfcv8nMEGsjAX6aE-SKDOceGqM2vAvXPhko,4407
-loqusdb/commands/load_profile.py,sha256=cflCbF9f77_HCH8xPnN8zSSocvIffRMnC2LPE0j7Xq8,3336
-loqusdb/commands/migrate.py,sha256=2C8YL-zVqnpnqg3JIyUr0rbVnb8-AGPVWNhicHnPKLo,667
-loqusdb/commands/restore.py,sha256=eqPX0yao0IAYS5SbjCdlsfSJRBbRByBLISUU2hTzqqs,1492
-loqusdb/commands/update.py,sha256=zz3wueaJVqJ1FKact-rpY2az__5oa1LnZKf7mgqNGPk,3211
-loqusdb/commands/view.py,sha256=zQag5kgvUFa8nW9OVte_qjit0n8wGLc3C3hwLOMGY6o,5111
-loqusdb/commands/wipe.py,sha256=WTOjyNooCUhtmZ6pdcPFa0PZrFc9E_pkLbnat_zP96M,553
-loqusdb/constants/__init__.py,sha256=r6y2TN8BqbKuh2Uyxq0trh-3A9xiWeStqWlvEPp-rSA,1645
-loqusdb/exceptions/__init__.py,sha256=Fq0UQg9TepWh19D7WT3dARyAHvorwJF6phhnZi2AkxE,88
-loqusdb/exceptions/case.py,sha256=n3mGF7RIc1imQFxnNJ1TWxeJeMWN4MHsKxoZb0m1-Os,92
-loqusdb/exceptions/profile.py,sha256=TVkRXh3ZbkNCmFHzZTCuhPP3iFWBwP1YQGD8IlSoCTo,98
-loqusdb/exceptions/vcf.py,sha256=QMpr9oRzYtMaHzP8wtSU5HiXGmi4k48YnjCilNZ0j2M,95
-loqusdb/models/__init__.py,sha256=yf0wONlDuGkztsOv15BFulYyQxdzqhuUKpL-R_clDVM,139
-loqusdb/models/case.py,sha256=EJOkrAMJfS6eID3E7QtWkoa_tMMf21KV3Z5B-U0c-Wk,1660
-loqusdb/models/identity.py,sha256=3DBlaZtrEtoiSU6nMXs7zY-mvy-9ew08-ZPjr_F3x3c,511
-loqusdb/models/profile_variant.py,sha256=7Y7HRnoOfhvThAuuaXWe2tOr0u2wktSFr5GmoLuDhg8,409
-loqusdb/models/variant.py,sha256=9QpM0ojJ24tKY9RrcgC411Wjnl3uRUawVYo6RWz1gXY,1068
-loqusdb/plugins/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
-loqusdb/plugins/mongo/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
-loqusdb/plugins/mongo/adapter.py,sha256=rRUQ2y_bZwjINbdTtfzJ_1JwE0Ns0usyAJ3o0O1Y10I,2593
-loqusdb/plugins/mongo/case.py,sha256=_LslGBB0F50IpjeEoV234JszC5TihAqFK3LBPMVD71k,3119
-loqusdb/plugins/mongo/profile_variant.py,sha256=madOBa3HacIN_T1V8mp7NEC4QOpqQBqSY6pKTObkMLA,600
-loqusdb/plugins/mongo/structural_variant.py,sha256=RPc3GF37-O3GqRS9Z5PRxsBXACljfRT4KFsrEl0jCu0,12955
-loqusdb/plugins/mongo/variant.py,sha256=NS3N8bhRap1b2ZSGjOw2LO6l7q_CzD00GZs499l38Tg,8715
-loqusdb/resources/__init__.py,sha256=JOx3Ppgtghx55cOJN0bfBonSXM5DM_v0OFdlT-Qybg4,639
-loqusdb/utils/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
-loqusdb/utils/annotate.py,sha256=cPNWlhsv6yoe3lxNfa9DytO5eACuM_mOJJw_mglVMN0,2646
-loqusdb/utils/case.py,sha256=aeTvyACJTDjzl-aOjAZaUzFMLisgFKMfcoXSvNAZz4s,2168
-loqusdb/utils/delete.py,sha256=-ddBM_QXKzlUN6egEJggKzXX1P-WEdi92HgaD1DJRtg,4843
-loqusdb/utils/load.py,sha256=rujUk89lL5jO35QfO8NKGJkjEJmN6PQxohKkxtpnw6Y,8252
-loqusdb/utils/migrate.py,sha256=9Q6kdIi9TpFVzDYptlEE8RqPPS5wyzfM3F8egzmmBBk,1113
-loqusdb/utils/profiling.py,sha256=3OizF7CpYvSl9kyl2g4KGJxbIRUqWfmfLxn3843XYDk,9164
-loqusdb/utils/update.py,sha256=1edJG-u24FgOSxyXAQEiyTG4IyK-Uo3lSIl5qyzcXsI,4433
-loqusdb/utils/variant.py,sha256=Lq5x9egVB-3rExBiRceF67TaL4Hp2gGoWMSRBEcnm4Q,2088
-loqusdb/utils/vcf.py,sha256=ybmrTBEPYa0FbUXo8ttlwATk13RnKjX9eIDbRDwCiVE,5175
-loqusdb-2.7.2.dist-info/LICENSE,sha256=urpFcJXw3elN9kV2fFutc-lXegjuu2lqP_GSy8_CAbs,1054
-loqusdb-2.7.2.dist-info/METADATA,sha256=NvWYugvrSe_lUtPGvWwXxUnqc-OiY0eJSRJ31gtatYk,6032
-loqusdb-2.7.2.dist-info/WHEEL,sha256=yQN5g4mg4AybRjkgi-9yy4iQEFibGQmlz78Pik5Or-A,92
-loqusdb-2.7.2.dist-info/entry_points.txt,sha256=39QklW01vy9ilBLcRPgCP18kN6oKXXLoOK50gZE7Jbs,59
-loqusdb-2.7.2.dist-info/top_level.txt,sha256=lRdRO6hqPhJEjFhfNsbCgVWMztvkYsjiOGK9DAL0UAI,8
-loqusdb-2.7.2.dist-info/RECORD,,

loqusdb-2.7.2.dist-info/entry_points.txt

@@ -1,3 +0,0 @@
-[console_scripts]
-loqusdb = loqusdb.__main__:base_command
-

loqusdb-2.7.2.dist-info/top_level.txt

@@ -1 +0,0 @@
-loqusdb