Skip to content

Study Splitter

gentropy.datasource.gwas_catalog.study_splitter.GWASCatalogStudySplitter

Splitting multi-trait GWAS Catalog studies.

Source code in src/gentropy/datasource/gwas_catalog/study_splitter.py 
 17
 18
 19
 20
 21
 22
 23
 24
 25
 26
 27
 28
 29
 30
 31
 32
 33
 34
 35
 36
 37
 38
 39
 40
 41
 42
 43
 44
 45
 46
 47
 48
 49
 50
 51
 52
 53
 54
 55
 56
 57
 58
 59
 60
 61
 62
 63
 64
 65
 66
 67
 68
 69
 70
 71
 72
 73
 74
 75
 76
 77
 78
 79
 80
 81
 82
 83
 84
 85
 86
 87
 88
 89
 90
 91
 92
 93
 94
 95
 96
 97
 98
 99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
class GWASCatalogStudySplitter:
    """Splitting multi-trait GWAS Catalog studies."""

    @staticmethod
    def _resolve_trait(
        study_trait: Column, association_trait: Column, p_value_text: Column
    ) -> Column:
        """Resolve trait names by consolidating association-level and study-level trait names.

        Args:
            study_trait (Column): Study-level trait name.
            association_trait (Column): Association-level trait name.
            p_value_text (Column): P-value text.

        Returns:
            Column: Resolved trait name.
        """
        return (
            f.when(
                (p_value_text.isNotNull()) & (p_value_text != ("no_pvalue_text")),
                f.concat(
                    association_trait,
                    f.lit(" ["),
                    p_value_text,
                    f.lit("]"),
                ),
            )
            .when(
                association_trait.isNotNull(),
                association_trait,
            )
            .otherwise(study_trait)
        )

    @staticmethod
    def _resolve_efo(association_efo: Column, study_efo: Column) -> Column:
        """Resolve EFOs by consolidating association-level and study-level EFOs.

        Args:
            association_efo (Column): EFO column from the association table.
            study_efo (Column): EFO column from the study table.

        Returns:
            Column: Consolidated EFO column.
        """
        return f.coalesce(f.split(association_efo, r"\/"), study_efo)

    @staticmethod
    def _resolve_study_id(study_id: Column, sub_study_description: Column) -> Column:
        """Resolve study IDs by exploding association-level information (e.g. pvalue_text, EFO).

        Args:
            study_id (Column): Study ID column.
            sub_study_description (Column): Sub-study description column from the association table.

        Returns:
            Column: Resolved study ID column.
        """
        split_w = Window.partitionBy(study_id).orderBy(sub_study_description)
        row_number = f.dense_rank().over(split_w)
        substudy_count = f.approx_count_distinct(row_number).over(split_w)
        return f.when(substudy_count == 1, study_id).otherwise(
            f.concat_ws("_", study_id, row_number)
        )

    @classmethod
    def split(
        cls: type[GWASCatalogStudySplitter],
        studies: StudyIndexGWASCatalog,
        associations: StudyLocusGWASCatalog,
    ) -> Tuple[StudyIndexGWASCatalog, StudyLocusGWASCatalog]:
        """Splitting multi-trait GWAS Catalog studies.

        If assigned disease of the study and the association don't agree, we assume the study needs to be split.
        Then disease EFOs, trait names and study ID are consolidated

        Args:
            studies (StudyIndexGWASCatalog): GWAS Catalog studies.
            associations (StudyLocusGWASCatalog): GWAS Catalog associations.

        Returns:
            Tuple[StudyIndexGWASCatalog, StudyLocusGWASCatalog]: Split studies and associations.
        """
        # Composite of studies and associations to resolve scattered information
        st_ass = (
            associations.df.join(f.broadcast(studies.df), on="studyId", how="inner")
            .select(
                "studyId",
                "subStudyDescription",
                cls._resolve_study_id(
                    f.col("studyId"), f.col("subStudyDescription")
                ).alias("updatedStudyId"),
                cls._resolve_trait(
                    f.col("traitFromSource"),
                    f.split("subStudyDescription", r"\|").getItem(0),
                    f.split("subStudyDescription", r"\|").getItem(1),
                ).alias("traitFromSource"),
                cls._resolve_efo(
                    f.split("subStudyDescription", r"\|").getItem(2),
                    f.col("traitFromSourceMappedIds"),
                ).alias("traitFromSourceMappedIds"),
            )
            .persist()
        )

        return (
            studies.update_study_id(
                st_ass.select(
                    "studyId",
                    "updatedStudyId",
                    "traitFromSource",
                    "traitFromSourceMappedIds",
                ).distinct()
            ),
            associations.update_study_id(
                st_ass.select(
                    "updatedStudyId", "studyId", "subStudyDescription"
                ).distinct()
            ).qc_ambiguous_study(),
        )

split(studies: StudyIndexGWASCatalog, associations: StudyLocusGWASCatalog) -> Tuple[StudyIndexGWASCatalog, StudyLocusGWASCatalog] classmethod

Splitting multi-trait GWAS Catalog studies.

If assigned disease of the study and the association don't agree, we assume the study needs to be split. Then disease EFOs, trait names and study ID are consolidated

Parameters:

Name Type Description Default
studies StudyIndexGWASCatalog

GWAS Catalog studies.

 required
associations StudyLocusGWASCatalog

GWAS Catalog associations.

 required

Returns:

Type Description
Tuple[StudyIndexGWASCatalog, StudyLocusGWASCatalog]

Tuple[StudyIndexGWASCatalog, StudyLocusGWASCatalog]: Split studies and associations.
Source code in src/gentropy/datasource/gwas_catalog/study_splitter.py 
 82
 83
 84
 85
 86
 87
 88
 89
 90
 91
 92
 93
 94
 95
 96
 97
 98
 99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
@classmethod
def split(
    cls: type[GWASCatalogStudySplitter],
    studies: StudyIndexGWASCatalog,
    associations: StudyLocusGWASCatalog,
) -> Tuple[StudyIndexGWASCatalog, StudyLocusGWASCatalog]:
    """Splitting multi-trait GWAS Catalog studies.

    If assigned disease of the study and the association don't agree, we assume the study needs to be split.
    Then disease EFOs, trait names and study ID are consolidated

    Args:
        studies (StudyIndexGWASCatalog): GWAS Catalog studies.
        associations (StudyLocusGWASCatalog): GWAS Catalog associations.

    Returns:
        Tuple[StudyIndexGWASCatalog, StudyLocusGWASCatalog]: Split studies and associations.
    """
    # Composite of studies and associations to resolve scattered information
    st_ass = (
        associations.df.join(f.broadcast(studies.df), on="studyId", how="inner")
        .select(
            "studyId",
            "subStudyDescription",
            cls._resolve_study_id(
                f.col("studyId"), f.col("subStudyDescription")
            ).alias("updatedStudyId"),
            cls._resolve_trait(
                f.col("traitFromSource"),
                f.split("subStudyDescription", r"\|").getItem(0),
                f.split("subStudyDescription", r"\|").getItem(1),
            ).alias("traitFromSource"),
            cls._resolve_efo(
                f.split("subStudyDescription", r"\|").getItem(2),
                f.col("traitFromSourceMappedIds"),
            ).alias("traitFromSourceMappedIds"),
        )
        .persist()
    )

    return (
        studies.update_study_id(
            st_ass.select(
                "studyId",
                "updatedStudyId",
                "traitFromSource",
                "traitFromSourceMappedIds",
            ).distinct()
        ),
        associations.update_study_id(
            st_ass.select(
                "updatedStudyId", "studyId", "subStudyDescription"
            ).distinct()
        ).qc_ambiguous_study(),
    )
  • 2023-09-25
  • 2024-01-18
  • Contributors
    • d0choa