@@ -546,7 +546,7 @@ class Test1000G2020Example:
546546 @pytest .fixture (scope = "class" )
547547 def ds (self , tmp_path_factory ):
548548 out = tmp_path_factory .mktemp ("data" ) / "example.vcf.zarr"
549- vcf2zarr .convert ([self .data_path ], out , worker_processes = 0 , local_alleles = True )
549+ vcf2zarr .convert ([self .data_path ], out , worker_processes = 0 )
550550 return sg .load_dataset (out )
551551
552552 def test_position (self , ds ):
@@ -617,33 +617,85 @@ def test_variant_MLEAC(self, ds):
617617 )
618618 nt .assert_array_equal (ds .variant_MLEAC .values , MLEAC )
619619
620- # def test_call_AD(self, ds):
621- # call_AD = [
622- # [[446, 0, -2, -2, -2], [393, 0, -2, -2, -2], [486, 0, -2, -2, -2]],
623- # [[446, 0, -2, -2, -2], [393, 0, -2, -2, -2], [486, 0, -2, -2, -2]],
624- # [[446, 0, -2, -2, -2], [393, 0, -2, -2, -2], [486, 0, -2, -2, -2]],
625- # [[446, 0, 0, -2, -2], [393, 0, 0, -2, -2], [486, 0, 0, -2, -2]],
626- # [[446, 0, 0, 0, -2], [393, 0, 0, 0, -2], [486, 0, 0, 0, -2]],
627- # [[446, 0, -2, -2, -2], [393, 0, -2, -2, -2], [486, 0, -2, -2, -2]],
628- # [[446, 0, -2, -2, -2], [393, 0, -2, -2, -2], [486, 0, -2, -2, -2]],
629- # [[446, 0, -2, -2, -2], [393, 0, -2, -2, -2], [486, 0, -2, -2, -2]],
630- # [[446, 0, -2, -2, -2], [393, 0, -2, -2, -2], [486, 0, -2, -2, -2]],
631- # [[446, 0, -2, -2, -2], [393, 0, -2, -2, -2], [486, 0, -2, -2, -2]],
632- # [[446, 0, -2, -2, -2], [393, 0, -2, -2, -2], [486, 0, -2, -2, -2]],
633- # [[446, 0, 0, 0, 0], [393, 0, 0, 0, 0], [486, 0, 0, 0, 0]],
634- # [[446, 0, 0, 0, 0], [393, 0, 0, 0, 0], [486, 0, 0, 0, 0]],
635- # [[446, 0, 0, 0, 0], [393, 0, 0, 0, 0], [486, 0, 0, 0, 0]],
636- # [[446, 0, 0, 0, -2], [393, 0, 0, 0, -2], [486, 0, 0, 0, -2]],
637- # [[446, 0, -2, -2, -2], [393, 0, -2, -2, -2], [486, 0, -2, -2, -2]],
638- # [[446, 0, -2, -2, -2], [393, 0, -2, -2, -2], [486, 0, -2, -2, -2]],
639- # [[446, 0, 0, -2, -2], [393, 0, 0, -2, -2], [486, 0, 0, -2, -2]],
640- # [[446, 0, -2, -2, -2], [393, 0, -2, -2, -2], [486, 0, -2, -2, -2]],
641- # [[446, 0, 0, -2, -2], [393, 0, 0, -2, -2], [486, 0, 0, -2, -2]],
642- # [[446, 0, 0, 0, -2], [393, 0, 0, 0, -2], [486, 0, 0, 0, -2]],
643- # [[446, 0, 0, 0, 0], [393, 0, 0, 0, 0], [486, 0, 0, 0, 0]],
644- # [[446, 0, 0, 0, -2], [393, 0, 0, 0, -2], [486, 0, 0, 0, -2]],
645- # ]
646- # nt.assert_array_equal(ds.call_AD.values, call_AD)
620+ def test_call_AD (self , ds ):
621+ call_AD = [
622+ [[446 , 0 , - 2 , - 2 , - 2 ], [393 , 0 , - 2 , - 2 , - 2 ], [486 , 0 , - 2 , - 2 , - 2 ]],
623+ [[446 , 0 , - 2 , - 2 , - 2 ], [393 , 0 , - 2 , - 2 , - 2 ], [486 , 0 , - 2 , - 2 , - 2 ]],
624+ [[446 , 0 , - 2 , - 2 , - 2 ], [393 , 0 , - 2 , - 2 , - 2 ], [486 , 0 , - 2 , - 2 , - 2 ]],
625+ [[446 , 0 , 0 , - 2 , - 2 ], [393 , 0 , 0 , - 2 , - 2 ], [486 , 0 , 0 , - 2 , - 2 ]],
626+ [[446 , 0 , 0 , 0 , - 2 ], [393 , 0 , 0 , 0 , - 2 ], [486 , 0 , 0 , 0 , - 2 ]],
627+ [[446 , 0 , - 2 , - 2 , - 2 ], [393 , 0 , - 2 , - 2 , - 2 ], [486 , 0 , - 2 , - 2 , - 2 ]],
628+ [[446 , 0 , - 2 , - 2 , - 2 ], [393 , 0 , - 2 , - 2 , - 2 ], [486 , 0 , - 2 , - 2 , - 2 ]],
629+ [[446 , 0 , - 2 , - 2 , - 2 ], [393 , 0 , - 2 , - 2 , - 2 ], [486 , 0 , - 2 , - 2 , - 2 ]],
630+ [[446 , 0 , - 2 , - 2 , - 2 ], [393 , 0 , - 2 , - 2 , - 2 ], [486 , 0 , - 2 , - 2 , - 2 ]],
631+ [[446 , 0 , - 2 , - 2 , - 2 ], [393 , 0 , - 2 , - 2 , - 2 ], [486 , 0 , - 2 , - 2 , - 2 ]],
632+ [[446 , 0 , - 2 , - 2 , - 2 ], [393 , 0 , - 2 , - 2 , - 2 ], [486 , 0 , - 2 , - 2 , - 2 ]],
633+ [[446 , 0 , 0 , 0 , 0 ], [393 , 0 , 0 , 0 , 0 ], [486 , 0 , 0 , 0 , 0 ]],
634+ [[446 , 0 , 0 , 0 , 0 ], [393 , 0 , 0 , 0 , 0 ], [486 , 0 , 0 , 0 , 0 ]],
635+ [[446 , 0 , 0 , 0 , 0 ], [393 , 0 , 0 , 0 , 0 ], [486 , 0 , 0 , 0 , 0 ]],
636+ [[446 , 0 , 0 , 0 , - 2 ], [393 , 0 , 0 , 0 , - 2 ], [486 , 0 , 0 , 0 , - 2 ]],
637+ [[446 , 0 , - 2 , - 2 , - 2 ], [393 , 0 , - 2 , - 2 , - 2 ], [486 , 0 , - 2 , - 2 , - 2 ]],
638+ [[446 , 0 , - 2 , - 2 , - 2 ], [393 , 0 , - 2 , - 2 , - 2 ], [486 , 0 , - 2 , - 2 , - 2 ]],
639+ [[446 , 0 , 0 , - 2 , - 2 ], [393 , 0 , 0 , - 2 , - 2 ], [486 , 0 , 0 , - 2 , - 2 ]],
640+ [[446 , 0 , - 2 , - 2 , - 2 ], [393 , 0 , - 2 , - 2 , - 2 ], [486 , 0 , - 2 , - 2 , - 2 ]],
641+ [[446 , 0 , 0 , - 2 , - 2 ], [393 , 0 , 0 , - 2 , - 2 ], [486 , 0 , 0 , - 2 , - 2 ]],
642+ [[446 , 0 , 0 , 0 , - 2 ], [393 , 0 , 0 , 0 , - 2 ], [486 , 0 , 0 , 0 , - 2 ]],
643+ [[446 , 0 , 0 , 0 , 0 ], [393 , 0 , 0 , 0 , 0 ], [486 , 0 , 0 , 0 , 0 ]],
644+ [[446 , 0 , 0 , 0 , - 2 ], [393 , 0 , 0 , 0 , - 2 ], [486 , 0 , 0 , 0 , - 2 ]],
645+ ]
646+ nt .assert_array_equal (ds .call_AD .values , call_AD )
647+
648+ def test_call_PID (self , ds ):
649+ call_PGT = ds ["call_PGT" ].values
650+ assert np .all (call_PGT == "." )
651+ assert call_PGT .shape == (23 , 3 )
652+
653+
654+ class Test1000G2020ExampleLocalAlleles :
655+ data_path = "tests/data/vcf/1kg_2020_chrM.vcf.gz"
656+
657+ @pytest .fixture (scope = "class" )
658+ def ds (self , tmp_path_factory ):
659+ out = tmp_path_factory .mktemp ("data" ) / "example.vcf.zarr"
660+ vcf2zarr .convert ([self .data_path ], out , worker_processes = 0 , local_alleles = True )
661+ return sg .load_dataset (out )
662+
663+ def test_position (self , ds ):
664+ # fmt: off
665+ pos = [
666+ 26 , 35 , 40 , 41 , 42 , 46 , 47 , 51 , 52 , 53 , 54 , 55 , 56 ,
667+ 57 , 58 , 59 , 60 , 61 , 62 , 63 , 64 , 65 , 66 ,
668+ ]
669+ # fmt: on
670+ nt .assert_array_equal (ds .variant_position .values , pos )
671+
672+ def test_alleles (self , ds ):
673+ alleles = [
674+ ["C" , "T" , "" , "" , "" ],
675+ ["G" , "A" , "" , "" , "" ],
676+ ["T" , "C" , "" , "" , "" ],
677+ ["C" , "T" , "CT" , "" , "" ],
678+ ["T" , "TC" , "C" , "TG" , "" ],
679+ ["T" , "C" , "" , "" , "" ],
680+ ["G" , "A" , "" , "" , "" ],
681+ ["T" , "C" , "" , "" , "" ],
682+ ["T" , "C" , "" , "" , "" ],
683+ ["G" , "A" , "" , "" , "" ],
684+ ["G" , "A" , "" , "" , "" ],
685+ ["TA" , "TAA" , "T" , "CA" , "AA" ],
686+ ["ATT" , "*" , "ATTT" , "ACTT" , "A" ],
687+ ["T" , "C" , "G" , "*" , "TC" ],
688+ ["T" , "A" , "C" , "*" , "" ],
689+ ["T" , "A" , "" , "" , "" ],
690+ ["T" , "A" , "" , "" , "" ],
691+ ["C" , "A" , "T" , "" , "" ],
692+ ["G" , "A" , "" , "" , "" ],
693+ ["T" , "C" , "A" , "" , "" ],
694+ ["C" , "T" , "CT" , "A" , "" ],
695+ ["TG" , "T" , "CG" , "TGG" , "TCGG" ],
696+ ["G" , "T" , "*" , "A" , "" ],
697+ ]
698+ nt .assert_array_equal (ds .variant_allele .values , alleles )
647699
648700 def test_call_LAD (self , ds ):
649701 call_LAD = [
@@ -682,11 +734,6 @@ def test_call_LAA(self, ds):
682734 # call_LPL = np.tile([0, -2, -2], (23, 3, 1))
683735 # nt.assert_array_equal(ds.call_LPL.values, call_LPL)
684736
685- def test_call_PID (self , ds ):
686- call_PGT = ds ["call_PGT" ].values
687- assert np .all (call_PGT == "." )
688- assert call_PGT .shape == (23 , 3 )
689-
690737
691738class Test1000G2020AnnotationsExample :
692739 data_path = "tests/data/vcf/1kg_2020_chr20_annotations.bcf"
0 commit comments