Skip to content

Commit

Permalink
tweak the test cases for new check_column_and_values_row_specific
Browse files Browse the repository at this point in the history
  • Loading branch information
@danlu1
danlu1 committed Nov 13, 2024
1 parent b8dc46f commit 49366dc
Showing 1 changed file with 43 additions and 27 deletions.
70 changes: 43 additions & 27 deletions tests/test_assay.py
View file
Original file line number Diff line number Diff line change
Expand Up @@ -10,7 +10,6 @@

GDC_DATA_DICT = {
"properties": {
"library_strategy": {"enum": ["value1", "value2"]},
"library_selection": {"enum": ["value1", "value2"]},
"platform": {"enum": ["value1", "value2"]},
"instrument_model": {"enum": ["value1", "value2"]},
Expand Down Expand Up @@ -170,7 +169,7 @@ def test__missingcols__validate(assay_info):

expected_warnings = (
"Assay_information.yaml: Doesn't have variant_classifications column. "
"This column will be added\n"
"This column will be added.\n"
"Assay_information.yaml: gene_padding is "
"by default 10 if not specified.\n"
)
Expand Down Expand Up @@ -262,28 +261,39 @@ def test_invalid__validate(assay_info):
"center abbreviation.\n"
"Assay_information.yaml: You are missing SEQ_ASSAY_IDs: SAGE-2\n"
"Assay_information.yaml: "
"Please double check your is_paired_end column. "
"This column must only be these values: True, False\n"
"Please double check your is_paired_end column. "
"Valid values are True, False. "
"You have 1 row(s) in your file where is_paired_end column contains invalid values. "
"The row(s) this occurs in are: [1]. Please correct.\n"
"Assay_information.yaml: "
"Please double check your library_selection column. "
"This column must only be these values: value1, value2\n"
"Please double check your library_selection column. "
"Valid values are value1, value2. "
"You have 2 row(s) in your file where library_selection column contains invalid values. "
"The row(s) this occurs in are: [0, 1]. Please correct.\n"
"Assay_information.yaml: "
"Please double check your library_strategy column. "
"This column must only be these values: Targeted Sequencing, WXS\n"
"Please double check your library_strategy column. "
"Valid values are Targeted Sequencing, WXS. "
"You have 1 row(s) in your file where library_strategy column contains invalid values. "
"The row(s) this occurs in are: [0]. Please correct.\n"
"Assay_information.yaml: "
"Please double check your platform column. "
"This column must only be these values: value1, value2\n"
"Please double check your platform column. "
"Valid values are value1, value2. "
"You have 2 row(s) in your file where platform column contains invalid values. "
"The row(s) this occurs in are: [0, 1]. Please correct.\n"
"Assay_information.yaml: "
"Please double check your instrument_model column. "
"This column must only be these values: value1, value2, "
"Illumina NovaSeq 6000, None\n"
"Please double check your instrument_model column. "
"Valid values are value1, value2, Illumina NovaSeq 6000, None. "
"You have 2 row(s) in your file where instrument_model column contains invalid values. "
"The row(s) this occurs in are: [0, 1]. Please correct.\n"
"Assay_information.yaml: "
"Please double check your variant_classifications column. "
"This column must only be these values: Splice_Site, "
"Please double check your variant_classifications column. "
"Valid values are Splice_Site, "
"Nonsense_Mutation, Frame_Shift_Del, Frame_Shift_Ins, "
"Nonstop_Mutation, Translation_Start_Site, In_Frame_Ins, "
"In_Frame_Del, Missense_Mutation, Intron, Splice_Region, "
"Silent, RNA, 5'UTR, 3'UTR, IGR, 5'Flank, 3'Flank, None\n"
"Silent, RNA, 5'UTR, 3'UTR, IGR, 5'Flank, 3'Flank, None. "
"You have 1 row(s) in your file where variant_classifications column contains invalid values. "
"The row(s) this occurs in are: [0]. Please correct.\n"
"Assay_information.yaml: "
"Please double check your read_length. "
"It must be an integer or null.\n"
Expand All @@ -294,23 +304,29 @@ def test_invalid__validate(assay_info):
"Please double check your gene_padding. "
"It must be an integer or blank.\n"
"Assay_information.yaml: "
"Please double check your calling_strategy column. This "
"column must only be these values: tumor_only, tumor_normal, "
"plasma_normal\n"
"Please double check your calling_strategy column. "
"Valid values are tumor_only, tumor_normal, plasma_normal. "
"You have 1 row(s) in your file where calling_strategy column contains invalid values. "
"The row(s) this occurs in are: [0]. Please correct.\n"
"Assay_information.yaml: "
"Please double check your specimen_tumor_cellularity. "
"It must in this format >(num)%. ie. >10%\n"
"Assay_information.yaml: "
"Please double check your alteration_types column. "
"This column must only be these values: snv, small_indels, "
"gene_level_cna, intragenic_cna, structural_variants\n"
"Please double check your alteration_types column. "
"Valid values are snv, small_indels, "
"gene_level_cna, intragenic_cna, structural_variants. "
"You have 1 row(s) in your file where alteration_types column contains invalid values. "
"The row(s) this occurs in are: [0]. Please correct.\n"
"Assay_information.yaml: "
"Please double check your preservation_technique column. "
"This column must only be these values: FFPE, fresh_frozen, NA\n"
"Please double check your preservation_technique column. "
"Valid values are FFPE, fresh_frozen, NA. "
"You have 1 row(s) in your file where preservation_technique column contains invalid values. "
"The row(s) this occurs in are: [0]. Please correct.\n"
"Assay_information.yaml: "
"Please double check your coverage column. "
"This column must only be these values: "
"hotspot_regions, coding_exons, introns, promoters\n"
"Please double check your coverage column. "
"Valid values are hotspot_regions, coding_exons, introns, promoters. "
"You have 1 row(s) in your file where coverage column contains invalid values. "
"The row(s) this occurs in are: [0]. Please correct.\n"
)

patch_get_gdc.called_once_with("read_group")
Expand Down

0 comments on commit 49366dc

Please sign in to comment.