ghga-transpiler 2.3.1__py3-none-any.whl → 3.0.0rc1__py3-none-any.whl
This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
- ghga_transpiler/cli.py +64 -36
- ghga_transpiler/config.py +136 -0
- ghga_transpiler/{config/exceptions.py → exceptions.py} +14 -2
- ghga_transpiler/metasheet_parser.py +125 -0
- ghga_transpiler/models.py +98 -0
- ghga_transpiler/transformations.py +1 -1
- ghga_transpiler/transpile.py +53 -0
- ghga_transpiler/{io.py → transpiler_io.py} +20 -21
- ghga_transpiler/workbook_parser.py +177 -0
- {ghga_transpiler-2.3.1.dist-info → ghga_transpiler-3.0.0rc1.dist-info}/METADATA +5 -6
- ghga_transpiler-3.0.0rc1.dist-info/RECORD +17 -0
- {ghga_transpiler-2.3.1.dist-info → ghga_transpiler-3.0.0rc1.dist-info}/WHEEL +1 -1
- ghga_transpiler/config/__init__.py +0 -20
- ghga_transpiler/config/config.py +0 -106
- ghga_transpiler/configs/0.10.yaml +0 -135
- ghga_transpiler/configs/1.0.yaml +0 -135
- ghga_transpiler/configs/1.1.yaml +0 -135
- ghga_transpiler/configs/2.0.yaml +0 -170
- ghga_transpiler/configs/2.1.yaml +0 -172
- ghga_transpiler/configs/__init__.py +0 -16
- ghga_transpiler/core.py +0 -155
- ghga_transpiler-2.3.1.dist-info/RECORD +0 -21
- {ghga_transpiler-2.3.1.dist-info → ghga_transpiler-3.0.0rc1.dist-info}/entry_points.txt +0 -0
- {ghga_transpiler-2.3.1.dist-info → ghga_transpiler-3.0.0rc1.dist-info}/licenses/LICENSE +0 -0
- {ghga_transpiler-2.3.1.dist-info → ghga_transpiler-3.0.0rc1.dist-info}/top_level.txt +0 -0
ghga_transpiler/configs/1.1.yaml
DELETED
|
@@ -1,135 +0,0 @@
|
|
|
1
|
-
ghga_metadata_version: 1.1.0
|
|
2
|
-
default_settings:
|
|
3
|
-
header_row: 1
|
|
4
|
-
start_row: 7
|
|
5
|
-
start_column: 1
|
|
6
|
-
transformations:
|
|
7
|
-
attributes: !!python/object/apply:ghga_transpiler.transformations.to_attributes []
|
|
8
|
-
format: !!python/object/apply:ghga_transpiler.transformations.to_snake_case []
|
|
9
|
-
forward_or_reverse: !!python/object/apply:ghga_transpiler.transformations.to_snake_case []
|
|
10
|
-
worksheets:
|
|
11
|
-
- settings:
|
|
12
|
-
end_column: 6
|
|
13
|
-
name: analyses
|
|
14
|
-
sheet_name: Analysis
|
|
15
|
-
- settings:
|
|
16
|
-
end_column: 9
|
|
17
|
-
name: analysis_process_output_files
|
|
18
|
-
sheet_name: AnalysisProcessOutputFile
|
|
19
|
-
- settings:
|
|
20
|
-
end_column: 5
|
|
21
|
-
name: analysis_processes
|
|
22
|
-
transformations:
|
|
23
|
-
study_input_files: !!python/object/apply:ghga_transpiler.transformations.to_list []
|
|
24
|
-
sample_input_files: !!python/object/apply:ghga_transpiler.transformations.to_list []
|
|
25
|
-
sequencing_process_input_files: !!python/object/apply:ghga_transpiler.transformations.to_list []
|
|
26
|
-
sheet_name: AnalysisProcess
|
|
27
|
-
- settings:
|
|
28
|
-
end_column: 10
|
|
29
|
-
name: biospecimens
|
|
30
|
-
transformations:
|
|
31
|
-
vital_status_at_sampling: !!python/object/apply:ghga_transpiler.transformations.to_snake_case []
|
|
32
|
-
age_at_sampling: !!python/object/apply:ghga_transpiler.transformations.to_snake_case []
|
|
33
|
-
sheet_name: Biospecimen
|
|
34
|
-
- settings:
|
|
35
|
-
end_column: 9
|
|
36
|
-
name: conditions
|
|
37
|
-
transformations:
|
|
38
|
-
disease_or_healthy: !!python/object/apply:ghga_transpiler.transformations.to_snake_case []
|
|
39
|
-
case_control_status: !!python/object/apply:ghga_transpiler.transformations.to_snake_case []
|
|
40
|
-
mutant_or_wildtype: !!python/object/apply:ghga_transpiler.transformations.to_snake_case []
|
|
41
|
-
sheet_name: Condition
|
|
42
|
-
- settings:
|
|
43
|
-
end_column: 3
|
|
44
|
-
name: data_access_committees
|
|
45
|
-
sheet_name: DataAccessCommittee
|
|
46
|
-
- settings:
|
|
47
|
-
end_column: 8
|
|
48
|
-
name: data_access_policies
|
|
49
|
-
transformations:
|
|
50
|
-
data_use_modifiers: !!python/object/apply:ghga_transpiler.transformations.to_snake_case_list []
|
|
51
|
-
data_use_permission: !!python/object/apply:ghga_transpiler.transformations.to_snake_case []
|
|
52
|
-
sheet_name: DataAccessPolicy
|
|
53
|
-
- settings:
|
|
54
|
-
end_column: 5
|
|
55
|
-
name: datasets
|
|
56
|
-
transformations:
|
|
57
|
-
types: !!python/object/apply:ghga_transpiler.transformations.to_list []
|
|
58
|
-
sheet_name: Dataset
|
|
59
|
-
- settings:
|
|
60
|
-
end_column: 6
|
|
61
|
-
name: individuals
|
|
62
|
-
transformations:
|
|
63
|
-
ancestries: !!python/object/apply:ghga_transpiler.transformations.to_list []
|
|
64
|
-
phenotypic_features: !!python/object/apply:ghga_transpiler.transformations.to_list []
|
|
65
|
-
karyotype: !!python/object/apply:ghga_transpiler.transformations.to_snake_case []
|
|
66
|
-
sex: !!python/object/apply:ghga_transpiler.transformations.to_snake_case []
|
|
67
|
-
sheet_name: Individual
|
|
68
|
-
- settings:
|
|
69
|
-
end_column: 14
|
|
70
|
-
name: library_preparation_protocols
|
|
71
|
-
transformations:
|
|
72
|
-
target_regions: !!python/object/apply:ghga_transpiler.transformations.to_list []
|
|
73
|
-
attributes: !!python/object/apply:ghga_transpiler.transformations.to_attributes []
|
|
74
|
-
library_layout: !!python/object/apply:ghga_transpiler.transformations.to_snake_case []
|
|
75
|
-
library_type: !!python/object/apply:ghga_transpiler.transformations.to_snake_case []
|
|
76
|
-
library_selection: !!python/object/apply:ghga_transpiler.transformations.to_snake_case []
|
|
77
|
-
library_preparation_kit_retail_name: !!python/object/apply:ghga_transpiler.transformations.to_snake_case []
|
|
78
|
-
rnaseq_strandedness: !!python/object/apply:ghga_transpiler.transformations.to_snake_case []
|
|
79
|
-
primer: !!python/object/apply:ghga_transpiler.transformations.to_snake_case []
|
|
80
|
-
end_bias: !!python/object/apply:ghga_transpiler.transformations.to_snake_case []
|
|
81
|
-
sheet_name: LibraryPreparationProtocol
|
|
82
|
-
- settings:
|
|
83
|
-
end_column: 9
|
|
84
|
-
name: publications
|
|
85
|
-
transformations:
|
|
86
|
-
xref: !!python/object/apply:ghga_transpiler.transformations.to_list []
|
|
87
|
-
sheet_name: Publication
|
|
88
|
-
- settings:
|
|
89
|
-
end_column: 9
|
|
90
|
-
name: sample_files
|
|
91
|
-
sheet_name: SampleFile
|
|
92
|
-
- settings:
|
|
93
|
-
end_column: 10
|
|
94
|
-
name: samples
|
|
95
|
-
transformations:
|
|
96
|
-
xref: !!python/object/apply:ghga_transpiler.transformations.to_list []
|
|
97
|
-
type: !!python/object/apply:ghga_transpiler.transformations.to_snake_case []
|
|
98
|
-
sheet_name: Sample
|
|
99
|
-
- settings:
|
|
100
|
-
end_column: 7
|
|
101
|
-
name: sequencing_experiments
|
|
102
|
-
sheet_name: SequencingExperiment
|
|
103
|
-
- settings:
|
|
104
|
-
end_column: 9
|
|
105
|
-
name: sequencing_process_files
|
|
106
|
-
sheet_name: SequencingProcessFile
|
|
107
|
-
- settings:
|
|
108
|
-
end_column: 12
|
|
109
|
-
name: sequencing_processes
|
|
110
|
-
sheet_name: SequencingProcess
|
|
111
|
-
- settings:
|
|
112
|
-
end_column: 17
|
|
113
|
-
name: sequencing_protocols
|
|
114
|
-
transformations:
|
|
115
|
-
instrument_model: !!python/object/apply:ghga_transpiler.transformations.to_snake_case []
|
|
116
|
-
flow_cell_type: !!python/object/apply:ghga_transpiler.transformations.to_snake_case []
|
|
117
|
-
umi_barcode_read: !!python/object/apply:ghga_transpiler.transformations.to_snake_case []
|
|
118
|
-
sample_barcode_read: !!python/object/apply:ghga_transpiler.transformations.to_snake_case []
|
|
119
|
-
sheet_name: SequencingProtocol
|
|
120
|
-
- settings:
|
|
121
|
-
end_column: 6
|
|
122
|
-
name: studies
|
|
123
|
-
transformations:
|
|
124
|
-
affiliations: !!python/object/apply:ghga_transpiler.transformations.to_list []
|
|
125
|
-
attributes: !!python/object/apply:ghga_transpiler.transformations.to_attributes []
|
|
126
|
-
type: !!python/object/apply:ghga_transpiler.transformations.to_snake_case []
|
|
127
|
-
sheet_name: Study
|
|
128
|
-
- settings:
|
|
129
|
-
end_column: 9
|
|
130
|
-
name: study_files
|
|
131
|
-
sheet_name: StudyFile
|
|
132
|
-
- settings:
|
|
133
|
-
end_column: 4
|
|
134
|
-
name: trios
|
|
135
|
-
sheet_name: Trio
|
ghga_transpiler/configs/2.0.yaml
DELETED
|
@@ -1,170 +0,0 @@
|
|
|
1
|
-
ghga_metadata_version: 2.0.0
|
|
2
|
-
default_settings:
|
|
3
|
-
header_row: 1
|
|
4
|
-
start_row: 7
|
|
5
|
-
start_column: 1
|
|
6
|
-
transformations:
|
|
7
|
-
attributes:
|
|
8
|
-
!!python/object/apply:ghga_transpiler.transformations.to_attributes []
|
|
9
|
-
format:
|
|
10
|
-
!!python/object/apply:ghga_transpiler.transformations.to_snake_case []
|
|
11
|
-
worksheets:
|
|
12
|
-
- settings:
|
|
13
|
-
end_column: 7
|
|
14
|
-
name: analyses
|
|
15
|
-
transformations:
|
|
16
|
-
research_data_files:
|
|
17
|
-
!!python/object/apply:ghga_transpiler.transformations.to_list []
|
|
18
|
-
sheet_name: Analysis
|
|
19
|
-
- settings:
|
|
20
|
-
end_column: 11
|
|
21
|
-
name: analysis_methods
|
|
22
|
-
transformations:
|
|
23
|
-
parameters:
|
|
24
|
-
!!python/object/apply:ghga_transpiler.transformations.to_attributes []
|
|
25
|
-
software_versions:
|
|
26
|
-
!!python/object/apply:ghga_transpiler.transformations.to_attributes []
|
|
27
|
-
sheet_name: AnalysisMethod
|
|
28
|
-
- settings:
|
|
29
|
-
end_column: 7
|
|
30
|
-
name: analysis_method_supporting_files
|
|
31
|
-
sheet_name: AnalysisMethodSupportingFile
|
|
32
|
-
- settings:
|
|
33
|
-
end_column: 7
|
|
34
|
-
name: studies
|
|
35
|
-
transformations:
|
|
36
|
-
affiliations:
|
|
37
|
-
!!python/object/apply:ghga_transpiler.transformations.to_list []
|
|
38
|
-
attributes:
|
|
39
|
-
!!python/object/apply:ghga_transpiler.transformations.to_attributes []
|
|
40
|
-
types:
|
|
41
|
-
!!python/object/apply:ghga_transpiler.transformations.to_snake_case_list []
|
|
42
|
-
attributes:
|
|
43
|
-
!!python/object/apply:ghga_transpiler.transformations.to_attributes []
|
|
44
|
-
sheet_name: Study
|
|
45
|
-
- settings:
|
|
46
|
-
end_column: 10
|
|
47
|
-
name: individuals
|
|
48
|
-
transformations:
|
|
49
|
-
phenotypic_features_terms:
|
|
50
|
-
!!python/object/apply:ghga_transpiler.transformations.to_list []
|
|
51
|
-
phenotypic_features_ids:
|
|
52
|
-
!!python/object/apply:ghga_transpiler.transformations.to_list []
|
|
53
|
-
diagnosis_ids:
|
|
54
|
-
!!python/object/apply:ghga_transpiler.transformations.to_list []
|
|
55
|
-
diagnosis_terms:
|
|
56
|
-
!!python/object/apply:ghga_transpiler.transformations.to_list []
|
|
57
|
-
sex:
|
|
58
|
-
!!python/object/apply:ghga_transpiler.transformations.to_snake_case []
|
|
59
|
-
ancestry_terms:
|
|
60
|
-
!!python/object/apply:ghga_transpiler.transformations.to_list []
|
|
61
|
-
ancestry_ids:
|
|
62
|
-
!!python/object/apply:ghga_transpiler.transformations.to_list []
|
|
63
|
-
sheet_name: Individual
|
|
64
|
-
- settings:
|
|
65
|
-
end_column: 7
|
|
66
|
-
name: individual_supporting_files
|
|
67
|
-
sheet_name: IndividualSupportingFile
|
|
68
|
-
- settings:
|
|
69
|
-
end_column: 21
|
|
70
|
-
name: samples
|
|
71
|
-
transformations:
|
|
72
|
-
type:
|
|
73
|
-
!!python/object/apply:ghga_transpiler.transformations.to_snake_case []
|
|
74
|
-
storage:
|
|
75
|
-
!!python/object/apply:ghga_transpiler.transformations.to_snake_case []
|
|
76
|
-
disease_or_healthy:
|
|
77
|
-
!!python/object/apply:ghga_transpiler.transformations.to_snake_case []
|
|
78
|
-
case_control_status:
|
|
79
|
-
!!python/object/apply:ghga_transpiler.transformations.to_snake_case []
|
|
80
|
-
xref: !!python/object/apply:ghga_transpiler.transformations.to_list []
|
|
81
|
-
biospecimen_age_at_sampling:
|
|
82
|
-
!!python/object/apply:ghga_transpiler.transformations.to_snake_case []
|
|
83
|
-
biospecimen_vital_status_at_sampling:
|
|
84
|
-
!!python/object/apply:ghga_transpiler.transformations.to_snake_case []
|
|
85
|
-
biospecimen_isolation:
|
|
86
|
-
!!python/object/apply:ghga_transpiler.transformations.to_snake_case []
|
|
87
|
-
biospecimen_storage:
|
|
88
|
-
!!python/object/apply:ghga_transpiler.transformations.to_snake_case []
|
|
89
|
-
attributes:
|
|
90
|
-
!!python/object/apply:ghga_transpiler.transformations.to_attributes []
|
|
91
|
-
sheet_name: Sample
|
|
92
|
-
- settings:
|
|
93
|
-
end_column: 23
|
|
94
|
-
name: experiment_methods
|
|
95
|
-
transformations:
|
|
96
|
-
library_type:
|
|
97
|
-
!!python/object/apply:ghga_transpiler.transformations.to_snake_case []
|
|
98
|
-
library_selection_methods:
|
|
99
|
-
!!python/object/apply:ghga_transpiler.transformations.to_snake_case_list []
|
|
100
|
-
library_preparation_kit_retail_name:
|
|
101
|
-
!!python/object/apply:ghga_transpiler.transformations.to_snake_case []
|
|
102
|
-
primer:
|
|
103
|
-
!!python/object/apply:ghga_transpiler.transformations.to_snake_case []
|
|
104
|
-
end_bias:
|
|
105
|
-
!!python/object/apply:ghga_transpiler.transformations.to_snake_case []
|
|
106
|
-
target_regions:
|
|
107
|
-
!!python/object/apply:ghga_transpiler.transformations.to_list []
|
|
108
|
-
rnaseq_strandedness:
|
|
109
|
-
!!python/object/apply:ghga_transpiler.transformations.to_snake_case []
|
|
110
|
-
instrument_model:
|
|
111
|
-
!!python/object/apply:ghga_transpiler.transformations.to_snake_case []
|
|
112
|
-
sequencing_layout:
|
|
113
|
-
!!python/object/apply:ghga_transpiler.transformations.to_snake_case []
|
|
114
|
-
flow_cell_type:
|
|
115
|
-
!!python/object/apply:ghga_transpiler.transformations.to_snake_case []
|
|
116
|
-
sample_barcode_read:
|
|
117
|
-
!!python/object/apply:ghga_transpiler.transformations.to_snake_case []
|
|
118
|
-
sequencing_layout:
|
|
119
|
-
!!python/object/apply:ghga_transpiler.transformations.to_snake_case []
|
|
120
|
-
attributes:
|
|
121
|
-
!!python/object/apply:ghga_transpiler.transformations.to_attributes []
|
|
122
|
-
sheet_name: ExperimentMethod
|
|
123
|
-
- settings:
|
|
124
|
-
end_column: 7
|
|
125
|
-
name: experiment_method_supporting_files
|
|
126
|
-
sheet_name: ExperimentMethodSupportingFile
|
|
127
|
-
- settings:
|
|
128
|
-
end_column: 8
|
|
129
|
-
name: experiments
|
|
130
|
-
sheet_name: Experiment
|
|
131
|
-
- settings:
|
|
132
|
-
end_column: 9
|
|
133
|
-
name: research_data_files
|
|
134
|
-
transformations:
|
|
135
|
-
experiments:
|
|
136
|
-
!!python/object/apply:ghga_transpiler.transformations.to_list []
|
|
137
|
-
format:
|
|
138
|
-
!!python/object/apply:ghga_transpiler.transformations.to_snake_case []
|
|
139
|
-
sheet_name: ResearchDataFile
|
|
140
|
-
- settings:
|
|
141
|
-
end_column: 7
|
|
142
|
-
name: process_data_files
|
|
143
|
-
sheet_name: ProcessDataFile
|
|
144
|
-
- settings:
|
|
145
|
-
end_column: 7
|
|
146
|
-
name: datasets
|
|
147
|
-
transformations:
|
|
148
|
-
types: !!python/object/apply:ghga_transpiler.transformations.to_list []
|
|
149
|
-
sheet_name: Dataset
|
|
150
|
-
- settings:
|
|
151
|
-
end_column: 11
|
|
152
|
-
name: data_access_policies
|
|
153
|
-
transformations:
|
|
154
|
-
data_use_modifier_terms:
|
|
155
|
-
!!python/object/apply:ghga_transpiler.transformations.to_snake_case_list []
|
|
156
|
-
data_use_permission_term:
|
|
157
|
-
!!python/object/apply:ghga_transpiler.transformations.to_snake_case []
|
|
158
|
-
data_use_modifier_ids:
|
|
159
|
-
!!python/object/apply:ghga_transpiler.transformations.to_list []
|
|
160
|
-
sheet_name: DataAccessPolicy
|
|
161
|
-
- settings:
|
|
162
|
-
end_column: 4
|
|
163
|
-
name: data_access_committees
|
|
164
|
-
sheet_name: DataAccessCommittee
|
|
165
|
-
- settings:
|
|
166
|
-
end_column: 9
|
|
167
|
-
name: publications
|
|
168
|
-
transformations:
|
|
169
|
-
xref: !!python/object/apply:ghga_transpiler.transformations.to_list []
|
|
170
|
-
sheet_name: Publication
|
ghga_transpiler/configs/2.1.yaml
DELETED
|
@@ -1,172 +0,0 @@
|
|
|
1
|
-
ghga_metadata_version: 2.1.0
|
|
2
|
-
default_settings:
|
|
3
|
-
header_row: 1
|
|
4
|
-
start_row: 7
|
|
5
|
-
start_column: 1
|
|
6
|
-
transformations:
|
|
7
|
-
attributes:
|
|
8
|
-
!!python/object/apply:ghga_transpiler.transformations.to_attributes []
|
|
9
|
-
format:
|
|
10
|
-
!!python/object/apply:ghga_transpiler.transformations.to_snake_case []
|
|
11
|
-
worksheets:
|
|
12
|
-
- settings:
|
|
13
|
-
end_column: 7
|
|
14
|
-
name: analyses
|
|
15
|
-
transformations:
|
|
16
|
-
research_data_files:
|
|
17
|
-
!!python/object/apply:ghga_transpiler.transformations.to_list []
|
|
18
|
-
sheet_name: Analysis
|
|
19
|
-
- settings:
|
|
20
|
-
end_column: 12
|
|
21
|
-
name: analysis_methods
|
|
22
|
-
transformations:
|
|
23
|
-
parameters:
|
|
24
|
-
!!python/object/apply:ghga_transpiler.transformations.to_attributes []
|
|
25
|
-
software_versions:
|
|
26
|
-
!!python/object/apply:ghga_transpiler.transformations.to_attributes []
|
|
27
|
-
attributes:
|
|
28
|
-
!!python/object/apply:ghga_transpiler.transformations.to_attributes []
|
|
29
|
-
sheet_name: AnalysisMethod
|
|
30
|
-
- settings:
|
|
31
|
-
end_column: 7
|
|
32
|
-
name: analysis_method_supporting_files
|
|
33
|
-
sheet_name: AnalysisMethodSupportingFile
|
|
34
|
-
- settings:
|
|
35
|
-
end_column: 7
|
|
36
|
-
name: studies
|
|
37
|
-
transformations:
|
|
38
|
-
affiliations:
|
|
39
|
-
!!python/object/apply:ghga_transpiler.transformations.to_list []
|
|
40
|
-
attributes:
|
|
41
|
-
!!python/object/apply:ghga_transpiler.transformations.to_attributes []
|
|
42
|
-
types:
|
|
43
|
-
!!python/object/apply:ghga_transpiler.transformations.to_snake_case_list []
|
|
44
|
-
attributes:
|
|
45
|
-
!!python/object/apply:ghga_transpiler.transformations.to_attributes []
|
|
46
|
-
sheet_name: Study
|
|
47
|
-
- settings:
|
|
48
|
-
end_column: 10
|
|
49
|
-
name: individuals
|
|
50
|
-
transformations:
|
|
51
|
-
phenotypic_features_terms:
|
|
52
|
-
!!python/object/apply:ghga_transpiler.transformations.to_list []
|
|
53
|
-
phenotypic_features_ids:
|
|
54
|
-
!!python/object/apply:ghga_transpiler.transformations.to_list []
|
|
55
|
-
diagnosis_ids:
|
|
56
|
-
!!python/object/apply:ghga_transpiler.transformations.to_list []
|
|
57
|
-
diagnosis_terms:
|
|
58
|
-
!!python/object/apply:ghga_transpiler.transformations.to_list []
|
|
59
|
-
sex:
|
|
60
|
-
!!python/object/apply:ghga_transpiler.transformations.to_snake_case []
|
|
61
|
-
ancestry_terms:
|
|
62
|
-
!!python/object/apply:ghga_transpiler.transformations.to_list []
|
|
63
|
-
ancestry_ids:
|
|
64
|
-
!!python/object/apply:ghga_transpiler.transformations.to_list []
|
|
65
|
-
sheet_name: Individual
|
|
66
|
-
- settings:
|
|
67
|
-
end_column: 7
|
|
68
|
-
name: individual_supporting_files
|
|
69
|
-
sheet_name: IndividualSupportingFile
|
|
70
|
-
- settings:
|
|
71
|
-
end_column: 21
|
|
72
|
-
name: samples
|
|
73
|
-
transformations:
|
|
74
|
-
type:
|
|
75
|
-
!!python/object/apply:ghga_transpiler.transformations.to_snake_case []
|
|
76
|
-
storage:
|
|
77
|
-
!!python/object/apply:ghga_transpiler.transformations.to_snake_case []
|
|
78
|
-
disease_or_healthy:
|
|
79
|
-
!!python/object/apply:ghga_transpiler.transformations.to_snake_case []
|
|
80
|
-
case_control_status:
|
|
81
|
-
!!python/object/apply:ghga_transpiler.transformations.to_snake_case []
|
|
82
|
-
xref: !!python/object/apply:ghga_transpiler.transformations.to_list []
|
|
83
|
-
biospecimen_age_at_sampling:
|
|
84
|
-
!!python/object/apply:ghga_transpiler.transformations.to_snake_case []
|
|
85
|
-
biospecimen_vital_status_at_sampling:
|
|
86
|
-
!!python/object/apply:ghga_transpiler.transformations.to_snake_case []
|
|
87
|
-
biospecimen_isolation:
|
|
88
|
-
!!python/object/apply:ghga_transpiler.transformations.to_snake_case []
|
|
89
|
-
biospecimen_storage:
|
|
90
|
-
!!python/object/apply:ghga_transpiler.transformations.to_snake_case []
|
|
91
|
-
attributes:
|
|
92
|
-
!!python/object/apply:ghga_transpiler.transformations.to_attributes []
|
|
93
|
-
sheet_name: Sample
|
|
94
|
-
- settings:
|
|
95
|
-
end_column: 23
|
|
96
|
-
name: experiment_methods
|
|
97
|
-
transformations:
|
|
98
|
-
library_type:
|
|
99
|
-
!!python/object/apply:ghga_transpiler.transformations.to_snake_case []
|
|
100
|
-
library_selection_methods:
|
|
101
|
-
!!python/object/apply:ghga_transpiler.transformations.to_snake_case_list []
|
|
102
|
-
library_preparation_kit_retail_name:
|
|
103
|
-
!!python/object/apply:ghga_transpiler.transformations.to_snake_case []
|
|
104
|
-
primer:
|
|
105
|
-
!!python/object/apply:ghga_transpiler.transformations.to_snake_case []
|
|
106
|
-
end_bias:
|
|
107
|
-
!!python/object/apply:ghga_transpiler.transformations.to_snake_case []
|
|
108
|
-
target_regions:
|
|
109
|
-
!!python/object/apply:ghga_transpiler.transformations.to_list []
|
|
110
|
-
rnaseq_strandedness:
|
|
111
|
-
!!python/object/apply:ghga_transpiler.transformations.to_snake_case []
|
|
112
|
-
instrument_model:
|
|
113
|
-
!!python/object/apply:ghga_transpiler.transformations.to_snake_case []
|
|
114
|
-
sequencing_layout:
|
|
115
|
-
!!python/object/apply:ghga_transpiler.transformations.to_snake_case []
|
|
116
|
-
flow_cell_type:
|
|
117
|
-
!!python/object/apply:ghga_transpiler.transformations.to_snake_case []
|
|
118
|
-
sample_barcode_read:
|
|
119
|
-
!!python/object/apply:ghga_transpiler.transformations.to_snake_case []
|
|
120
|
-
sequencing_layout:
|
|
121
|
-
!!python/object/apply:ghga_transpiler.transformations.to_snake_case []
|
|
122
|
-
attributes:
|
|
123
|
-
!!python/object/apply:ghga_transpiler.transformations.to_attributes []
|
|
124
|
-
sheet_name: ExperimentMethod
|
|
125
|
-
- settings:
|
|
126
|
-
end_column: 7
|
|
127
|
-
name: experiment_method_supporting_files
|
|
128
|
-
sheet_name: ExperimentMethodSupportingFile
|
|
129
|
-
- settings:
|
|
130
|
-
end_column: 8
|
|
131
|
-
name: experiments
|
|
132
|
-
sheet_name: Experiment
|
|
133
|
-
- settings:
|
|
134
|
-
end_column: 9
|
|
135
|
-
name: research_data_files
|
|
136
|
-
transformations:
|
|
137
|
-
experiments:
|
|
138
|
-
!!python/object/apply:ghga_transpiler.transformations.to_list []
|
|
139
|
-
format:
|
|
140
|
-
!!python/object/apply:ghga_transpiler.transformations.to_snake_case []
|
|
141
|
-
sheet_name: ResearchDataFile
|
|
142
|
-
- settings:
|
|
143
|
-
end_column: 7
|
|
144
|
-
name: process_data_files
|
|
145
|
-
sheet_name: ProcessDataFile
|
|
146
|
-
- settings:
|
|
147
|
-
end_column: 7
|
|
148
|
-
name: datasets
|
|
149
|
-
transformations:
|
|
150
|
-
types: !!python/object/apply:ghga_transpiler.transformations.to_list []
|
|
151
|
-
sheet_name: Dataset
|
|
152
|
-
- settings:
|
|
153
|
-
end_column: 11
|
|
154
|
-
name: data_access_policies
|
|
155
|
-
transformations:
|
|
156
|
-
data_use_modifier_terms:
|
|
157
|
-
!!python/object/apply:ghga_transpiler.transformations.to_snake_case_list []
|
|
158
|
-
data_use_permission_term:
|
|
159
|
-
!!python/object/apply:ghga_transpiler.transformations.to_snake_case []
|
|
160
|
-
data_use_modifier_ids:
|
|
161
|
-
!!python/object/apply:ghga_transpiler.transformations.to_list []
|
|
162
|
-
sheet_name: DataAccessPolicy
|
|
163
|
-
- settings:
|
|
164
|
-
end_column: 4
|
|
165
|
-
name: data_access_committees
|
|
166
|
-
sheet_name: DataAccessCommittee
|
|
167
|
-
- settings:
|
|
168
|
-
end_column: 9
|
|
169
|
-
name: publications
|
|
170
|
-
transformations:
|
|
171
|
-
xref: !!python/object/apply:ghga_transpiler.transformations.to_list []
|
|
172
|
-
sheet_name: Publication
|
|
@@ -1,16 +0,0 @@
|
|
|
1
|
-
# Copyright 2021 - 2025 Universität Tübingen, DKFZ, EMBL, and Universität zu Köln
|
|
2
|
-
# for the German Human Genome-Phenome Archive (GHGA)
|
|
3
|
-
|
|
4
|
-
# Licensed under the Apache License, Version 2.0 (the "License");
|
|
5
|
-
# you may not use this file except in compliance with the License.
|
|
6
|
-
# You may obtain a copy of the License at
|
|
7
|
-
|
|
8
|
-
# http://www.apache.org/licenses/LICENSE-2.0
|
|
9
|
-
|
|
10
|
-
# Unless required by applicable law or agreed to in writing, software
|
|
11
|
-
# distributed under the License is distributed on an "AS IS" BASIS,
|
|
12
|
-
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
|
|
13
|
-
# See the License for the specific language governing permissions and
|
|
14
|
-
# limitations under the License.
|
|
15
|
-
#
|
|
16
|
-
#
|
ghga_transpiler/core.py
DELETED
|
@@ -1,155 +0,0 @@
|
|
|
1
|
-
# Copyright 2021 - 2025 Universität Tübingen, DKFZ, EMBL, and Universität zu Köln
|
|
2
|
-
# for the German Human Genome-Phenome Archive (GHGA)
|
|
3
|
-
|
|
4
|
-
# Licensed under the Apache License, Version 2.0 (the "License");
|
|
5
|
-
# you may not use this file except in compliance with the License.
|
|
6
|
-
# You may obtain a copy of the License at
|
|
7
|
-
|
|
8
|
-
# http://www.apache.org/licenses/LICENSE-2.0
|
|
9
|
-
|
|
10
|
-
# Unless required by applicable law or agreed to in writing, software
|
|
11
|
-
# distributed under the License is distributed on an "AS IS" BASIS,
|
|
12
|
-
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
|
|
13
|
-
# See the License for the specific language governing permissions and
|
|
14
|
-
# limitations under the License.
|
|
15
|
-
#
|
|
16
|
-
#
|
|
17
|
-
|
|
18
|
-
"""This module contains functionalities for processing excel sheets into json object."""
|
|
19
|
-
|
|
20
|
-
from __future__ import annotations
|
|
21
|
-
|
|
22
|
-
from collections.abc import Callable
|
|
23
|
-
from importlib import resources
|
|
24
|
-
|
|
25
|
-
import semver
|
|
26
|
-
from openpyxl import Workbook
|
|
27
|
-
|
|
28
|
-
from . import config
|
|
29
|
-
|
|
30
|
-
|
|
31
|
-
class InvalidSematicVersion(Exception):
|
|
32
|
-
"""Raised when a version string is invalid."""
|
|
33
|
-
|
|
34
|
-
|
|
35
|
-
class GHGAWorkbook:
|
|
36
|
-
"""A GHGA metadata XLSX workbook"""
|
|
37
|
-
|
|
38
|
-
def __init__(self, workbook: Workbook, configs_package: resources.Package):
|
|
39
|
-
"""Create a new GHGAWorkbook object from an XLSX workbook"""
|
|
40
|
-
self.workbook = workbook
|
|
41
|
-
self.wb_version = GHGAWorkbook._get_version(workbook)
|
|
42
|
-
self.config = config.load_config(self.major_minor_version, configs_package)
|
|
43
|
-
|
|
44
|
-
@staticmethod
|
|
45
|
-
def _get_version(workbook):
|
|
46
|
-
"""Function to get workbook version from the worksheet _properties"""
|
|
47
|
-
if "__properties" in workbook.sheetnames:
|
|
48
|
-
try:
|
|
49
|
-
return semver.Version.parse(workbook["__properties"].cell(1, 1).value)
|
|
50
|
-
except ValueError:
|
|
51
|
-
raise InvalidSematicVersion(
|
|
52
|
-
"Unable to extract metadata model version from the provided workbook (not a valid semantic version)."
|
|
53
|
-
) from None
|
|
54
|
-
raise SyntaxError(
|
|
55
|
-
"Unable to extract metadata model version from the provided workbook (missing)."
|
|
56
|
-
)
|
|
57
|
-
|
|
58
|
-
@property
|
|
59
|
-
def major_minor_version(self):
|
|
60
|
-
"""Returns only major and minor version numbers"""
|
|
61
|
-
return f"{self.wb_version.major}.{self.wb_version.minor}"
|
|
62
|
-
|
|
63
|
-
|
|
64
|
-
def get_worksheet_rows(
|
|
65
|
-
worksheet,
|
|
66
|
-
min_row: int | None,
|
|
67
|
-
max_row: int,
|
|
68
|
-
min_col: int | None,
|
|
69
|
-
max_col: int | None,
|
|
70
|
-
) -> list:
|
|
71
|
-
"""Function to create a list of rows of a worksheet"""
|
|
72
|
-
return list(
|
|
73
|
-
row
|
|
74
|
-
for row in worksheet.iter_rows(
|
|
75
|
-
min_row, max_row, min_col, max_col, values_only=True
|
|
76
|
-
)
|
|
77
|
-
if not all(cell is None for cell in row)
|
|
78
|
-
)
|
|
79
|
-
|
|
80
|
-
|
|
81
|
-
def get_header(
|
|
82
|
-
worksheet,
|
|
83
|
-
header_row: int | None,
|
|
84
|
-
min_col: int | None,
|
|
85
|
-
max_col: int | None,
|
|
86
|
-
) -> list[str]:
|
|
87
|
-
"""Function to return a list column names of a worksheet"""
|
|
88
|
-
return list(
|
|
89
|
-
cell.value
|
|
90
|
-
for row in worksheet.iter_rows(header_row, header_row, min_col, max_col)
|
|
91
|
-
for cell in row
|
|
92
|
-
)
|
|
93
|
-
|
|
94
|
-
|
|
95
|
-
def convert_rows(header, rows) -> list[dict]:
|
|
96
|
-
"""Function to return list of dictionaries, rows as worksheet row values and
|
|
97
|
-
column names as keys
|
|
98
|
-
"""
|
|
99
|
-
return [
|
|
100
|
-
{
|
|
101
|
-
key: value
|
|
102
|
-
for key, value in zip(header, row)
|
|
103
|
-
if value is not None and value != ""
|
|
104
|
-
}
|
|
105
|
-
for row in rows
|
|
106
|
-
]
|
|
107
|
-
|
|
108
|
-
|
|
109
|
-
def transform_rows(
|
|
110
|
-
rows: list[dict], transformations: dict[str, Callable] | None
|
|
111
|
-
) -> list[dict]:
|
|
112
|
-
"""Transforms row values if it is applicable with a given function"""
|
|
113
|
-
transformed = []
|
|
114
|
-
for row in rows:
|
|
115
|
-
transformed_row = {}
|
|
116
|
-
for key, value in row.items():
|
|
117
|
-
if transformations and key in transformations:
|
|
118
|
-
transformed_row[key] = transformations[key](value)
|
|
119
|
-
else:
|
|
120
|
-
transformed_row[key] = value
|
|
121
|
-
transformed.append(transformed_row)
|
|
122
|
-
return transformed
|
|
123
|
-
|
|
124
|
-
|
|
125
|
-
def convert_workbook(ghga_workbook: GHGAWorkbook) -> dict:
|
|
126
|
-
"""Function to convert an input spreadsheet into JSON"""
|
|
127
|
-
converted_workbook = {}
|
|
128
|
-
for sheet in ghga_workbook.config.worksheets:
|
|
129
|
-
if sheet.settings is not None:
|
|
130
|
-
if sheet.sheet_name in ghga_workbook.workbook:
|
|
131
|
-
rows = get_worksheet_rows(
|
|
132
|
-
ghga_workbook.workbook[sheet.sheet_name],
|
|
133
|
-
sheet.settings.start_row,
|
|
134
|
-
ghga_workbook.workbook[sheet.sheet_name].max_row,
|
|
135
|
-
sheet.settings.start_column,
|
|
136
|
-
sheet.settings.end_column,
|
|
137
|
-
)
|
|
138
|
-
|
|
139
|
-
header = get_header(
|
|
140
|
-
ghga_workbook.workbook[sheet.sheet_name],
|
|
141
|
-
sheet.settings.header_row,
|
|
142
|
-
sheet.settings.start_column,
|
|
143
|
-
sheet.settings.end_column,
|
|
144
|
-
)
|
|
145
|
-
converted_rows = convert_rows(header, rows)
|
|
146
|
-
transformed_rows = transform_rows(
|
|
147
|
-
converted_rows, sheet.settings.transformations
|
|
148
|
-
)
|
|
149
|
-
converted_workbook[sheet.settings.name] = transformed_rows
|
|
150
|
-
else:
|
|
151
|
-
converted_workbook[sheet.settings.name] = []
|
|
152
|
-
|
|
153
|
-
else:
|
|
154
|
-
raise ValueError(f"{sheet.settings} will never be None")
|
|
155
|
-
return converted_workbook
|
|
@@ -1,21 +0,0 @@
|
|
|
1
|
-
ghga_transpiler/__init__.py,sha256=b_JlBjNvngOiIolsfKKbwonyfOtS7b5_LJKLuB5LyNQ,908
|
|
2
|
-
ghga_transpiler/__main__.py,sha256=k82ZsGuAOTrbqB4rCNkGOS-AWU8GICtD1_owfiey-Sk,847
|
|
3
|
-
ghga_transpiler/cli.py,sha256=CvRQdiprRj1ZHjXgHpZ7JhDoAum7WB-sdFyMKknjtxk,2522
|
|
4
|
-
ghga_transpiler/core.py,sha256=Fhf5pICDmVPKrUOhQNo6uN-mZB8mA1Vlyq3VGeeVSZ8,5235
|
|
5
|
-
ghga_transpiler/io.py,sha256=e9DT5l5UumX2NcBJOkrR0y7ho_FPvK914WeFRZ4rOjw,1805
|
|
6
|
-
ghga_transpiler/transformations.py,sha256=th320v3teZwjn13T-xS9n2_lCrBo5Wrl9wq0qOnbMEs,2222
|
|
7
|
-
ghga_transpiler/config/__init__.py,sha256=CsnuJcqYMBx5zDcMgAzuKGRs5GQSS9N0evzDZ-T_g7k,817
|
|
8
|
-
ghga_transpiler/config/config.py,sha256=YlIMQwnnpi8BelrdHtbZpJsp9RfwHCDZ56tCvU89wts,3823
|
|
9
|
-
ghga_transpiler/config/exceptions.py,sha256=WB0YM5hJAn7iVGB_SY0p6Q4D0do1ti7AM4pKnMyVfLc,1090
|
|
10
|
-
ghga_transpiler/configs/0.10.yaml,sha256=hsS8xwzJ00SRSb-VIfM-oM8UM0hOdR3xeuflf1fra4Y,5683
|
|
11
|
-
ghga_transpiler/configs/1.0.yaml,sha256=MgSEkDpZjn7I76uLFAfpCfmOqOaCBjEojn6VDX6TnUc,5683
|
|
12
|
-
ghga_transpiler/configs/1.1.yaml,sha256=rgxTV7bEVMwk1UeJ8hAeCZcqTwX10KXQKoaYnpV0bxI,5683
|
|
13
|
-
ghga_transpiler/configs/2.0.yaml,sha256=3lwXFjtbEW-mWGj99d3PaWJflLnuy7WAS5AvNubDJlI,6637
|
|
14
|
-
ghga_transpiler/configs/2.1.yaml,sha256=A-vJbxUcW0yyAKpgz2IkN81n92wxIC2A3sYEq-UZ9d0,6738
|
|
15
|
-
ghga_transpiler/configs/__init__.py,sha256=xL0pfHoTdBh0LhdqA8lpdhp3W68HJVXF7AURhBM7pU4,686
|
|
16
|
-
ghga_transpiler-2.3.1.dist-info/licenses/LICENSE,sha256=nKgIWA3zoWy18_YSW6e9Gd6_p7d7ccwjNZrfJzvKALE,11452
|
|
17
|
-
ghga_transpiler-2.3.1.dist-info/METADATA,sha256=RKXPB_H8N5uLyKugDxk2ptbUG0jxsV4xLaVijDh3C8c,4452
|
|
18
|
-
ghga_transpiler-2.3.1.dist-info/WHEEL,sha256=Nw36Djuh_5VDukK0H78QzOX-_FQEo6V37m3nkm96gtU,91
|
|
19
|
-
ghga_transpiler-2.3.1.dist-info/entry_points.txt,sha256=Fr_VQJynZZl95NXjrQW2gE0bCocgjKVNEZQpaLls8po,65
|
|
20
|
-
ghga_transpiler-2.3.1.dist-info/top_level.txt,sha256=TksRpDO3Y4mvK-B09UhQU3ceZpG19fljNlVDGTfjg8o,16
|
|
21
|
-
ghga_transpiler-2.3.1.dist-info/RECORD,,
|
|
File without changes
|
|
File without changes
|
|
File without changes
|