|
| 1 | +import pytest |
| 2 | +from cyvcf2 import VCF |
| 3 | + |
| 4 | +from bio2zarr.csi import read_csi |
| 5 | +from bio2zarr.tbi import read_tabix |
| 6 | +from bio2zarr.vcf_partition import get_csi_path |
| 7 | +from bio2zarr.vcf_partition import get_tabix_path |
| 8 | + |
| 9 | +from .utils import count_variants, path_for_test |
| 10 | + |
| 11 | + |
| 12 | +class TestCsiIndex: |
| 13 | + @pytest.mark.parametrize( |
| 14 | + "vcf_file", |
| 15 | + [ |
| 16 | + "CEUTrio.20.21.gatk3.4.csi.g.vcf.bgz", |
| 17 | + ], |
| 18 | + ) |
| 19 | + def test_record_counts(self, shared_datadir, vcf_file): |
| 20 | + # Check record counts in csi with actual count of VCF |
| 21 | + vcf_path = path_for_test(shared_datadir, vcf_file, True) |
| 22 | + csi_path = get_csi_path(vcf_path) |
| 23 | + assert csi_path is not None |
| 24 | + csi = read_csi(csi_path) |
| 25 | + |
| 26 | + for i, contig in enumerate(VCF(vcf_path).seqnames): |
| 27 | + assert csi.record_counts[i] == count_variants(vcf_path, contig) |
| 28 | + |
| 29 | + @pytest.mark.parametrize( |
| 30 | + "file", |
| 31 | + ["CEUTrio.20.21.gatk3.4.g.vcf.bgz", "CEUTrio.20.21.gatk3.4.g.vcf.bgz.tbi"], |
| 32 | + ) |
| 33 | + def test_invalid_csi(self, shared_datadir, file): |
| 34 | + with pytest.raises(ValueError, match=r"File not in CSI format."): |
| 35 | + read_csi(path_for_test(shared_datadir, file, True)) |
| 36 | + |
| 37 | + @pytest.mark.parametrize( |
| 38 | + "file", |
| 39 | + ["CEUTrio.20.21.gatk3.4.g.vcf.bgz", "CEUTrio.20.21.gatk3.4.g.vcf.bgz.tbi"], |
| 40 | + ) |
| 41 | + def test_invalid_csi(self, shared_datadir, file): |
| 42 | + with pytest.raises(ValueError, match=r"File not in CSI format."): |
| 43 | + read_csi(path_for_test(shared_datadir, file, True)) |
| 44 | + |
| 45 | + |
| 46 | +class TestTabixIndex: |
| 47 | + @pytest.mark.parametrize( |
| 48 | + "vcf_file", |
| 49 | + [ |
| 50 | + "CEUTrio.20.21.gatk3.4.g.vcf.bgz", |
| 51 | + ], |
| 52 | + ) |
| 53 | + def test_record_counts(self, shared_datadir, vcf_file): |
| 54 | + # Check record counts in tabix with actual count of VCF |
| 55 | + vcf_path = path_for_test(shared_datadir, vcf_file, True) |
| 56 | + tabix_path = get_tabix_path(vcf_path) |
| 57 | + assert tabix_path is not None |
| 58 | + tabix = read_tabix(tabix_path) |
| 59 | + |
| 60 | + for i, contig in enumerate(tabix.sequence_names): |
| 61 | + assert tabix.record_counts[i] == count_variants(vcf_path, contig) |
| 62 | + |
| 63 | + @pytest.mark.parametrize( |
| 64 | + "file", |
| 65 | + ["CEUTrio.20.21.gatk3.4.g.vcf.bgz", "CEUTrio.20.21.gatk3.4.csi.g.vcf.bgz.csi"], |
| 66 | + ) |
| 67 | + def test_read_tabix__invalid_tbi(self, shared_datadir, file): |
| 68 | + with pytest.raises(ValueError, match=r"File not in Tabix format."): |
| 69 | + read_tabix(path_for_test(shared_datadir, file, True)) |
0 commit comments