Skip to content

Variants

gentropy.datasource.gnomad.variants.GnomADVariants

GnomAD variants included in the GnomAD genomes dataset.

Source code in src/gentropy/datasource/gnomad/variants.py 
 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
137
138
139
140
141
142
143
144
145
146
class GnomADVariants:
    """GnomAD variants included in the GnomAD genomes dataset."""

    def __init__(
        self,
        gnomad_genomes_path: str = GnomadVariantConfig().gnomad_genomes_path,
        gnomad_variant_populations: list[
            VariantPopulation | str
        ] = GnomadVariantConfig().gnomad_variant_populations,
        hash_threshold: int = VariantIndexConfig().hash_threshold,
    ):
        """Initialize.

        Args:
            gnomad_genomes_path (str): Path to gnomAD genomes hail table.
            gnomad_variant_populations (list[VariantPopulation | str]): List of populations to include.
            hash_threshold (int): longer variant ids will be hashed.

        All defaults are stored in GnomadVariantConfig.
        """
        self.gnomad_genomes_path = gnomad_genomes_path
        self.gnomad_variant_populations = gnomad_variant_populations
        self.lenght_threshold = hash_threshold

    def as_variant_index(self: GnomADVariants) -> VariantIndex:
        """Generate variant annotation dataset from gnomAD.

        Some relevant modifications to the original dataset are:

        1. The transcript consequences features provided by VEP are filtered to only refer to the Ensembl canonical transcript.
        2. Genome coordinates are liftovered from GRCh38 to GRCh37 to keep as annotation.
        3. Field names are converted to camel case to follow the convention.

        Returns:
            VariantIndex: GnomaAD variants dataset.
        """
        # Load variants dataset
        ht = hl.read_table(
            self.gnomad_genomes_path,
            _load_refs=False,
        )

        # Drop non biallelic variants
        ht = ht.filter(ht.alleles.length() == 2)

        # Select relevant fields and nested records to create class
        return VariantIndex(
            _df=(
                ht.select(
                    # Extract mandatory fields:
                    variantId=hl.str("_").join(
                        [
                            ht.locus.contig.replace("chr", ""),
                            hl.str(ht.locus.position),
                            ht.alleles[0],
                            ht.alleles[1],
                        ]
                    ),
                    chromosome=ht.locus.contig.replace("chr", ""),
                    position=ht.locus.position,
                    referenceAllele=ht.alleles[0],
                    alternateAllele=ht.alleles[1],
                    # Extract allele frequencies from populations of interest:
                    alleleFrequencies=hl.set(
                        [f"{pop}_adj" for pop in self.gnomad_variant_populations]
                    ).map(
                        lambda p: hl.struct(
                            populationName=p,
                            alleleFrequency=ht.freq[ht.globals.freq_index_dict[p]].AF,
                        )
                    ),
                    # Extract in silico predictors:
                    inSilicoPredictors=hl.array(
                        [
                            hl.struct(
                                method=hl.str("spliceai"),
                                assessment=hl.missing(hl.tstr),
                                score=hl.expr.functions.float32(
                                    ht.in_silico_predictors.spliceai_ds_max
                                ),
                                assessmentFlag=hl.missing(hl.tstr),
                                targetId=hl.missing(hl.tstr),
                            ),
                            hl.struct(
                                method=hl.str("pangolin"),
                                assessment=hl.missing(hl.tstr),
                                score=hl.expr.functions.float32(
                                    ht.in_silico_predictors.pangolin_largest_ds
                                ),
                                assessmentFlag=hl.missing(hl.tstr),
                                targetId=hl.missing(hl.tstr),
                            ),
                        ]
                    ),
                    # Extract cross references to GnomAD:
                    dbXrefs=hl.array(
                        [
                            hl.struct(
                                id=hl.str("-").join(
                                    [
                                        ht.locus.contig.replace("chr", ""),
                                        hl.str(ht.locus.position),
                                        ht.alleles[0],
                                        ht.alleles[1],
                                    ]
                                ),
                                source=hl.str("gnomad"),
                            )
                        ]
                    ),
                )
                .key_by("chromosome", "position")
                .drop("locus", "alleles")
                .select_globals()
                .to_spark(flatten=False)
                .withColumn(
                    "variantId",
                    VariantIndex.hash_long_variant_ids(
                        f.col("variantId"),
                        f.col("chromosome"),
                        f.col("position"),
                        self.lenght_threshold,
                    ),
                )
                .withColumn("mostSevereConsequenceId", f.lit(None).cast(t.StringType()))
            ),
            _schema=VariantIndex.get_schema(),
        )

__init__(gnomad_genomes_path: str = GnomadVariantConfig().gnomad_genomes_path, gnomad_variant_populations: list[VariantPopulation | str] = GnomadVariantConfig().gnomad_variant_populations, hash_threshold: int = VariantIndexConfig().hash_threshold)

Initialize.

Parameters:

Name Type Description Default
gnomad_genomes_path str

Path to gnomAD genomes hail table.

 gnomad_genomes_path
gnomad_variant_populations list[VariantPopulation | str]

List of populations to include.

 gnomad_variant_populations
hash_threshold int

longer variant ids will be hashed.

 hash_threshold

All defaults are stored in GnomadVariantConfig.

Source code in src/gentropy/datasource/gnomad/variants.py 
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
def __init__(
    self,
    gnomad_genomes_path: str = GnomadVariantConfig().gnomad_genomes_path,
    gnomad_variant_populations: list[
        VariantPopulation | str
    ] = GnomadVariantConfig().gnomad_variant_populations,
    hash_threshold: int = VariantIndexConfig().hash_threshold,
):
    """Initialize.

    Args:
        gnomad_genomes_path (str): Path to gnomAD genomes hail table.
        gnomad_variant_populations (list[VariantPopulation | str]): List of populations to include.
        hash_threshold (int): longer variant ids will be hashed.

    All defaults are stored in GnomadVariantConfig.
    """
    self.gnomad_genomes_path = gnomad_genomes_path
    self.gnomad_variant_populations = gnomad_variant_populations
    self.lenght_threshold = hash_threshold

as_variant_index() -> VariantIndex

Generate variant annotation dataset from gnomAD.

Some relevant modifications to the original dataset are:

  1. The transcript consequences features provided by VEP are filtered to only refer to the Ensembl canonical transcript.
  2. Genome coordinates are liftovered from GRCh38 to GRCh37 to keep as annotation.
  3. Field names are converted to camel case to follow the convention.

Returns:

Name Type Description
VariantIndex VariantIndex

GnomaAD variants dataset.
Source code in src/gentropy/datasource/gnomad/variants.py 
 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
137
138
139
140
141
142
143
144
145
146
def as_variant_index(self: GnomADVariants) -> VariantIndex:
    """Generate variant annotation dataset from gnomAD.

    Some relevant modifications to the original dataset are:

    1. The transcript consequences features provided by VEP are filtered to only refer to the Ensembl canonical transcript.
    2. Genome coordinates are liftovered from GRCh38 to GRCh37 to keep as annotation.
    3. Field names are converted to camel case to follow the convention.

    Returns:
        VariantIndex: GnomaAD variants dataset.
    """
    # Load variants dataset
    ht = hl.read_table(
        self.gnomad_genomes_path,
        _load_refs=False,
    )

    # Drop non biallelic variants
    ht = ht.filter(ht.alleles.length() == 2)

    # Select relevant fields and nested records to create class
    return VariantIndex(
        _df=(
            ht.select(
                # Extract mandatory fields:
                variantId=hl.str("_").join(
                    [
                        ht.locus.contig.replace("chr", ""),
                        hl.str(ht.locus.position),
                        ht.alleles[0],
                        ht.alleles[1],
                    ]
                ),
                chromosome=ht.locus.contig.replace("chr", ""),
                position=ht.locus.position,
                referenceAllele=ht.alleles[0],
                alternateAllele=ht.alleles[1],
                # Extract allele frequencies from populations of interest:
                alleleFrequencies=hl.set(
                    [f"{pop}_adj" for pop in self.gnomad_variant_populations]
                ).map(
                    lambda p: hl.struct(
                        populationName=p,
                        alleleFrequency=ht.freq[ht.globals.freq_index_dict[p]].AF,
                    )
                ),
                # Extract in silico predictors:
                inSilicoPredictors=hl.array(
                    [
                        hl.struct(
                            method=hl.str("spliceai"),
                            assessment=hl.missing(hl.tstr),
                            score=hl.expr.functions.float32(
                                ht.in_silico_predictors.spliceai_ds_max
                            ),
                            assessmentFlag=hl.missing(hl.tstr),
                            targetId=hl.missing(hl.tstr),
                        ),
                        hl.struct(
                            method=hl.str("pangolin"),
                            assessment=hl.missing(hl.tstr),
                            score=hl.expr.functions.float32(
                                ht.in_silico_predictors.pangolin_largest_ds
                            ),
                            assessmentFlag=hl.missing(hl.tstr),
                            targetId=hl.missing(hl.tstr),
                        ),
                    ]
                ),
                # Extract cross references to GnomAD:
                dbXrefs=hl.array(
                    [
                        hl.struct(
                            id=hl.str("-").join(
                                [
                                    ht.locus.contig.replace("chr", ""),
                                    hl.str(ht.locus.position),
                                    ht.alleles[0],
                                    ht.alleles[1],
                                ]
                            ),
                            source=hl.str("gnomad"),
                        )
                    ]
                ),
            )
            .key_by("chromosome", "position")
            .drop("locus", "alleles")
            .select_globals()
            .to_spark(flatten=False)
            .withColumn(
                "variantId",
                VariantIndex.hash_long_variant_ids(
                    f.col("variantId"),
                    f.col("chromosome"),
                    f.col("position"),
                    self.lenght_threshold,
                ),
            )
            .withColumn("mostSevereConsequenceId", f.lit(None).cast(t.StringType()))
        ),
        _schema=VariantIndex.get_schema(),
    )
  • 2023-09-25
  • 2024-01-18
  • Contributors
    • d0choa