Type: | Package |
Title: | Access Open Target |
Version: | 2.0.0 |
Date: | 2025-07-08 |
Maintainer: | Amir Feizi <afeizi@gmail.com> |
Description: | Interact seamlessly with Open Target GraphQL endpoint to query and retrieve tidy data tables, facilitating the analysis of gene, disease, drug, and genetic data. For more information about the Open Target API (https://platform.opentargets.org/api). |
Depends: | R (≥ 3.1.0) |
Imports: | ghql, cli, dplyr, jsonlite, magrittr, tibble, tidyr, httr |
License: | MIT + file LICENSE |
Encoding: | UTF-8 |
Suggests: | testthat (≥ 3.0.0) |
Config/testthat/edition: | 3 |
RoxygenNote: | 7.3.2 |
URL: | https://amirfeizi.github.io/otargen/ |
BugReports: | https://github.com/amirfeizi/otargen/issues |
NeedsCompilation: | no |
Packaged: | 2025-07-15 18:45:05 UTC; amir_ |
Author: | Amir Feizi [aut, cre] |
Repository: | CRAN |
Date/Publication: | 2025-07-15 19:10:02 UTC |
Retrieve Adverse Events data for a specified drug.
Description
This function queries the Open Targets GraphQL API to retrieve adverse events data for a specified drug.
Usage
adverseEventsQuery(chemblId, index = 0, size = 10)
Arguments
chemblId |
Character: ChEMBL ID of the target drug (e.g., "CHEMBL1016"). |
index |
Integer: Page index for pagination (default: 0). |
size |
Integer: Number of records to retrieve (default: 10). |
Value
Returns a tibble containing adverse events data for the specified drug.
Examples
## Not run:
result <- adverseEventsQuery(chemblId = "CHEMBL1016", size = 10)
result <- adverseEventsQuery(chemblId = "CHEMBL1016", index = 0, size = 10)
## End(Not run)
Retrieve ChEMBL data for a specified gene and disease.
Description
This function queries the Open Targets GraphQL API to retrieve ChEMBL evidence data for a specified gene and disease.
Usage
chemblQuery(ensemblId, efoId, cursor = NULL, size = 10)
Arguments
ensemblId |
Character: ENSEMBL ID of the target gene (e.g., "ENSG00000080815"). |
efoId |
Character: EFO ID of the target disease (e.g., "MONDO_0004975"). |
cursor |
Character: Cursor for pagination (default: NULL). |
size |
Integer: Number of records to retrieve (default: 10). |
Value
Returns a tibble containing ChEMBL evidence data for the specified gene and disease.
Examples
## Not run:
result <- chemblQuery(ensemblId = "ENSG00000080815", efoId =
"MONDO_0004975",
size = 10)
result <- chemblQuery(ensemblId = "ENSG00000080815", efoId =
"MONDO_0004975",
cursor = NULL, size = 10)
## End(Not run)
Retrieve ClinVar data for a specified gene and disease.
Description
This function queries the Open Targets GraphQL API to retrieve ClinVar evidence data for a specified gene and disease.
Usage
clinVarQuery(ensemblId, efoId, cursor = NULL, size = 10)
Arguments
ensemblId |
Character: ENSEMBL ID of the target gene (e.g., "ENSG00000080815"). |
efoId |
Character: EFO ID of the target disease (e.g., "MONDO_0004975"). |
cursor |
Character: Cursor for pagination (default: NULL). |
size |
Integer: Number of records to retrieve (default: 10). |
Value
Returns a tibble containing ClinVar evidence data for the specified gene and disease.
Examples
## Not run:
result <- clinVarQuery(ensemblId = "ENSG00000080815", efoId =
"MONDO_0004975", size = 10)
result <- clinVarQuery(ensemblId = "ENSG00000080815", efoId =
"MONDO_0004975", cursor = NULL, size = 10)
## End(Not run)
Retrieve Comparative Genomics data for a specified gene.
Description
This function queries the Open Targets GraphQL API to retrieve comparative genomics data for a specified gene.
Usage
compGenomicsQuery(ensemblId)
Arguments
ensemblId |
Character: ENSEMBL ID of the target gene (e.g., ENSG00000169174). |
Value
Returns a data frame containing comparative genomics data for the specified gene.
Examples
## Not run:
result <- compGenomicsQuery(ensemblId = "ENSG00000169174")
## End(Not run)
Retrieve DepMap Essentiality data for a specified gene.
Description
This function queries the Open Targets GraphQL API to retrieve DepMap essentiality data for a specified gene.
Usage
depMapQuery(ensgId)
Arguments
ensgId |
Character: ENSEMBL ID of the target gene (e.g., ENSG00000141510). |
Value
Returns a tibble containing DepMap essentiality data for the specified gene.
Examples
## Not run:
result <- depMapQuery(ensgId = "ENSG00000141510")
## End(Not run)
Retrieve Europe PMC data for a specified gene and disease.
Description
This function queries the Open Targets GraphQL API to retrieve Europe PMC evidence data for a specified gene and disease.
Usage
europePMCQuery(ensemblId, efoId, cursor = NULL, size = 50)
Arguments
ensemblId |
Character: ENSEMBL ID of the target gene (e.g., "ENSG00000080815"). |
efoId |
Character: EFO ID of the target disease (e.g., "MONDO_0004975"). |
cursor |
Character: Cursor for pagination (default: NULL). |
size |
Integer: Number of records to retrieve (default: 50). |
Value
Returns a tibble containing Europe PMC evidence data for the specified gene and disease.
Examples
## Not run:
result <- europePMCQuery(ensemblId = "ENSG00000080815",
efoId = "MONDO_0004975", size = 50)
result <- europePMCQuery(ensemblId = "ENSG00000080815",
efoId = "MONDO_0004975", cursor = NULL, size = 50)
## End(Not run)
Retrieve Gene Burden data for a specified gene and disease.
Description
This function queries the Open Targets GraphQL API to retrieve gene burden evidence data for a specified gene and disease.
Usage
geneBurdenQuery(ensemblId, efoId, size = 3500)
Arguments
ensemblId |
Character: ENSEMBL ID of the target gene (e.g., "ENSG00000137642"). |
efoId |
Character: EFO ID of the target disease (e.g., "MONDO_0004975"). |
size |
Integer: Number of records to retrieve (default: 3500). |
Value
Returns a tibble containing gene burden evidence data for the specified gene and disease.
Examples
## Not run:
result <- geneBurdenQuery(ensemblId = "ENSG00000137642", efoId =
"MONDO_0004975", size = 3500)
## End(Not run)
Retrieve Gene Ontology data for a specified gene.
Description
This function queries the Open Targets GraphQL API to retrieve gene ontology data for a specified gene.
Usage
geneOntologyQuery(ensgId)
Arguments
ensgId |
Character: ENSEMBL ID of the target gene (e.g., ENSG00000141510). |
Value
Returns a tibble containing gene ontology data for the specified gene.
Examples
## Not run:
result <- geneOntologyQuery(ensgId = "ENSG00000141510")
## End(Not run)
Retrieve Genetic Constraint data for a specified gene.
Description
This function queries the Open Targets Platform GraphQL API to retrieve genetic constraint data for a specified gene, such as pLI or LOEUF scores.
Usage
geneticConstraintQuery(ensgId)
Arguments
ensgId |
Character: ENSEMBL ID of the target gene (e.g., "ENSG00000141510"). |
Value
Returns a tibble containing genetic constraint data for the specified gene.
Examples
## Not run:
result <- geneticConstraintQuery(ensgId = "ENSG00000141510")
## End(Not run)
Retrieve Genomics England data for a specified gene and disease.
Description
This function queries the Open Targets GraphQL API to retrieve Genomics England evidence data for a specified gene and disease.
Usage
genomicsEnglandQuery(ensemblId, efoId, size = 3500)
Arguments
ensemblId |
Character: ENSEMBL ID of the target gene (e.g., "ENSG00000080815"). |
efoId |
Character: EFO ID of the target disease (e.g., "MONDO_0004975"). |
size |
Integer: Number of records to retrieve (default: 3500). |
Value
Returns a tibble containing Genomics England evidence data for the specified gene and disease.
Examples
## Not run:
result <- genomicsEnglandQuery(ensemblId = "ENSG00000080815", efoId =
"MONDO_0004975", size = 3500)
## End(Not run)
Retrieve calculated GWAS colocalisation data
Description
This function retrieves colocalisation data for a specific study locus from a GWAS study with other GWAS studies. It returns a data frame of the studies that colocalise with the input study locus, including details on the study, reported trait, index variant, and calculated colocalisation method outputs.
Usage
gwasColocalisation(study_locus_id, size = 500, index = 0)
Arguments
study_locus_id |
Character: Open Target Genetics generated ID for the study locus (e.g., "5a86bfd40d2ebecf6ce97bbe8a737512"). |
size |
Integer: Number of rows to fetch per page. Default: 500. |
index |
Integer: Page index for pagination. Default: 0. |
Value
Returns a data frame of the studies that colocalise with the input study locus. The table consists of the following data structure:
study.studyId
: Character vector. Study identifier.study.traitReported
: Character vector. Reported trait associated with the colocalisation.study.projectId
: Character vector. Project identifier for the study.study.publicationFirstAuthor
: Character vector. First author of the publication.indexVariant.id
: Character vector. Index variant identifier.indexVariant.position
: Integer vector. Index variant position.indexVariant.chromosome
: Character vector. Index variant chromosome.indexVariant.referenceAllele
: Character vector. Reference allele of the variant.indexVariant.alternateAllele
: Character vector. Alternate allele of the variant.pValueMantissa
: Numeric vector. Mantissa of the p-value for the colocalisation.pValueExponent
: Integer vector. Exponent of the p-value for the colocalisation.numberColocalisingVariants
: Integer vector. Number of colocalising variants.colocalisationMethod
: Character vector. Method used for colocalisation analysis.h3
: Numeric vector. H3 value associated with the colocalisation.h4
: Numeric vector. H4 value associated with the colocalisation.clpp
: Numeric vector. Colocalisation posterior probability.betaRatioSignAverage
: Numeric vector. Average sign of the beta ratio.
References
Giambartolomei, Claudia et al. “Bayesian test for colocalisation between pairs of genetic association studies using summary statistics.” PLoS genetics vol. 10,5 e1004383. 15 May. 2014, doi:10.1371/journal.pgen.1004383
Examples
## Not run:
colocalisation_data <- gwasColocalisation(study_locus_id = "5a86bfd40d2ebecf6ce97bbe8a737512",
size = 500, index = 0)
## End(Not run)
Retrieve GWAS credible set data
Description
Provided with a study ID and a lead variant ID, this function returns a data frame consisting of all the associated credible set tag variants with the corresponding statistical data.
Usage
gwasCredibleSet(study_id, variant_id)
Arguments
study_id |
Character: Study ID(s) generated by Open Targets (e.g GCST90002357). |
variant_id |
Character: generated ID for variants by Open Targets (e.g. 1_154119580_C_A) or rsId (rs2494663). |
Value
Returns a data frame of results from the credible set of variants for a specific lead variant with the following columns:
tagVariant.id
: Data frame. A table of IDs of the tag variant.tagVariant.rsId
: Character vector. rsID of the tag variant.beta
: Numeric. Beta value.postProb
: Numeric. Posterior probability.pval
: Numeric. P-value.se
: Numeric. Standard error.MultisignalMethod
: Character vector. Multisignal method.logABF
: Numeric. Logarithm of approximate Bayes factor.is95
: Logical. Indicates if the variant has a 95is99
: Logical. Indicates if the variant has a 99
Examples
## Not run:
result <- gwasCredibleSet(study_id="GCST90002357",
variant_id="1_154119580_C_A")
result <- gwasCredibleSet(study_id="GCST90002357", variant_id="rs2494663")
## End(Not run)
Retrieve GWAS Credible Sets data for a specified target and disease.
Description
This function queries the Open Targets Platform GraphQL API to retrieve GWAS credible sets evidence data for a specified target gene and disease.
Usage
gwasCredibleSetsQuery(ensemblId, efoId, size = 500)
Arguments
ensemblId |
Character. Ensembl gene ID, e.g., "ENSG00000169174". |
efoId |
Character. EFO disease ID, e.g., "EFO_0004911". |
size |
Integer. Number of rows to fetch. Default: 500. |
Value
A tibble with credible set evidence or NULL if no data found.
Examples
## Not run:
result <- gwasCredibleSetsQuery(
ensemblId = "ENSG00000169174",
efoId = "EFO_0004911",
size = 5
)
print(result)
## End(Not run)
Retrieve Cancer Hallmarks data for a specified gene.
Description
This function queries the Open Targets GraphQL API to retrieve cancer hallmarks data for a specified gene.
Usage
hallmarksQuery(ensgId)
Arguments
ensgId |
Character: ENSEMBL ID of the target gene (e.g., ENSG00000141510). |
Value
Returns a tibble containing cancer hallmarks data for the specified gene.
Examples
## Not run:
result <- hallmarksQuery(ensgId = "ENSG00000141510")
## End(Not run)
Retrieve Indications data for a specified drug.
Description
This function queries the Open Targets GraphQL API to retrieve indications data for a specified drug.
Usage
indicationsQuery(chemblId)
Arguments
chemblId |
Character: ChEMBL ID of the target drug (e.g., "CHEMBL1016"). |
Value
Returns a tibble containing indications data for the specified drug.
Examples
## Not run:
result <- indicationsQuery(chemblId = "CHEMBL1016")
## End(Not run)
Retrieve Interactions data for a specified gene.
Description
This function queries the Open Targets GraphQL API to retrieve molecular interaction data for a specified gene.
Usage
interactionsQuery(ensgId, sourceDatabase = NULL, index = 0, size = 10)
Arguments
ensgId |
Character: ENSEMBL ID of the target gene (e.g., ENSG00000141510). |
sourceDatabase |
Character: Source database for interactions (e.g., "intact") (default: NULL). |
index |
Integer: Page index for pagination (default: 0). |
size |
Integer: Number of records to retrieve (default: 10). |
Value
Returns a tibble containing interactions data for the specified gene.
Examples
## Not run:
result <- interactionsQuery(ensgId = "ENSG00000141510",
sourceDatabase = "intact", size = 10)
result <- interactionsQuery(ensgId = "ENSG00000141510",
sourceDatabase = "intact", index = 0, size = 10)
## End(Not run)
Retrieve Known Drugs data for a specified drug.
Description
This function queries the Open Targets GraphQL API to retrieve known drugs data for a specified drug.
Usage
knownDrugsChemblQuery(chemblId, cursor = NULL, freeTextQuery = NULL, size = 10)
Arguments
chemblId |
Character: ChEMBL ID of the target drug (e.g., "CHEMBL1016"). |
cursor |
Character: Cursor for pagination (default: NULL). |
freeTextQuery |
Character: Free text query to filter results (default: NULL). |
size |
Integer: Number of records to retrieve (default: 10). |
Value
Returns a tibble containing known drugs data for the specified drug.
Examples
## Not run:
result <- knownDrugsChemblQuery(chemblId = "CHEMBL1016", size = 10)
result <- knownDrugsChemblQuery(chemblId = "CHEMBL1016", cursor = NULL,
freeTextQuery = NULL, size = 10)
## End(Not run)
Retrieve Known Drugs data for a specified gene.
Description
This function queries the Open Targets GraphQL API to retrieve known drugs data for a specified gene.
Usage
knownDrugsGeneQuery(ensgId, cursor = NULL, freeTextQuery = NULL, size = 10)
Arguments
ensgId |
Character: ENSEMBL ID of the target gene (e.g., ENSG00000169174). |
cursor |
Character: Cursor for pagination (default: NULL). |
freeTextQuery |
Character: Free text query to filter results (default: NULL). |
size |
Integer: Number of records to retrieve (default: 10). |
Value
Returns a data frame containing known drugs data for the specified gene.
Examples
## Not run:
result <- knownDrugsGeneQuery(ensgId = "ENSG00000169174", size = 10)
result <- knownDrugsGeneQuery(ensgId = "ENSG00000169174",
cursor = NULL, freeTextQuery = NULL, size = 10)
## End(Not run)
Retrieve Locus-to-Gene Predictions data for a specified study locus.
Description
This function queries the Open Targets GraphQL API to retrieve locus-to-gene prediction data for a specified study locus.
Usage
locus2GeneQuery(studyLocusId)
Arguments
studyLocusId |
Character: ID of the target study locus (e.g., "fa375739ca2a6b825ce5cc69d117e84b"). |
Value
Returns a tibble containing locus-to-gene prediction data for the specified study locus.
Examples
## Not run:
result <- locus2GeneQuery(studyLocusId = "fa375739ca2a6b825ce5cc69d117e84b")
## End(Not run)
Retrieve Mechanisms of Action data for a specified drug.
Description
This function queries the Open Targets GraphQL API to retrieve mechanisms of action data for a specified drug.
Usage
mechanismsOfActionQuery(chemblId)
Arguments
chemblId |
Character: ChEMBL ID of the target drug (e.g., "CHEMBL1016"). |
Value
Returns a tibble containing mechanisms of action data for the specified drug.
Examples
## Not run:
result <- mechanismsOfActionQuery(chemblId = "CHEMBL1016")
## End(Not run)
Retrieve Mouse Phenotypes data for a specified gene.
Description
This function queries the Open Targets GraphQL API to retrieve mouse phenotypes data for a specified gene.
Usage
mousePhenotypesQuery(ensemblId)
Arguments
ensemblId |
Character: ENSEMBL ID of the target gene (e.g., ENSG00000169174). |
Value
Returns a data frame containing mouse phenotypes data for the specified gene.
Examples
## Not run:
result <- mousePhenotypesQuery(ensemblId = "ENSG00000169174")
## End(Not run)
Retrieve Orphanet data for a specified gene and disease.
Description
This function queries the Open Targets GraphQL API to retrieve Orphanet evidence data for a specified gene and disease.
Usage
orphanetQuery(ensemblId, efoId, size = 3500)
Arguments
ensemblId |
Character: ENSEMBL ID of the target gene (e.g., "ENSG00000080815"). |
efoId |
Character: EFO ID of the target disease (e.g., "MONDO_0004975"). |
size |
Integer: Number of records to retrieve (default: 3500). |
Value
Returns a tibble containing Orphanet evidence data for the specified gene and disease.
Examples
## Not run:
result <- orphanetQuery(ensemblId = "ENSG00000080815", efoId =
"MONDO_0004975", size = 3500)
## End(Not run)
Retrieves overlap info for a study and a list of studies
Description
For an input study ID and a list of other study IDs, this function returns two elements. One contains the overlap information in a table format, and the other element is the variant intersection set, representing an overlap between two variants of the two given studies.
Usage
overlapInfoForStudy(study_id, study_ids = list())
Arguments
study_id |
Character: Study ID(s) generated by Open Targets (e.g GCST90002357). |
study_ids |
Character: generated ID for variants by Open Targets (e.g. 1_154119580_C_A) or rsId (rs2494663). |
Value
A list containing a data frame of overlap information and the variant intersection set. The overlap information table (overlap_info) consists of the following columns:
studyId
: Character vector. Study ID.traitReported
: Character vector. Reported trait.traitCategory
: Character vector. Trait category.variantIdA
: Character vector. Variant ID from study A.variantIdB
: Character vector. Variant ID from study B.overlapAB
: Integer vector. Number of overlaps between variants A and B.distinctA
: Integer vector. Number of distinct variants in study A.distinctB
: Integer vector. Number of distinct variants in study B.study.studyId
: Character vector. Study ID from study list.study.traitReported
: Character vector. Reported trait from study list.study.traitCategory
: Character vector. Trait category from study list.
The variant intersection set (variant_intersection_set) is a character vector representing the intersection of variants.
Examples
## Not run:
result <- overlapInfoForStudy(study_id = "GCST90002357",
study_ids = list("GCST90025975", "GCST90025962"))
## End(Not run)
Retrieve Pathways data for a specified gene.
Description
This function queries the Open Targets GraphQL API to retrieve pathways data for a specified gene.
Usage
pathwaysQuery(ensgId)
Arguments
ensgId |
Character: ENSEMBL ID of the target gene (e.g., ENSG00000105397). |
Value
Returns a tibble containing pathways data for the specified gene.
Examples
## Not run:
result <- pathwaysQuery(ensgId = "ENSG00000105397")
## End(Not run)
Retrieve Pharmacogenomics data for a specified drug.
Description
This function queries the Open Targets GraphQL API to retrieve pharmacogenomics data for a specified drug.
Usage
pharmacogenomicsChemblQuery(chemblId)
Arguments
chemblId |
Character: ChEMBL ID of the target drug (e.g., "CHEMBL1016"). |
Value
Returns a tibble containing pharmacogenomics data for the specified drug.
Examples
## Not run:
result <- pharmacogenomicsChemblQuery(chemblId = "CHEMBL1016")
## End(Not run)
Retrieve Pharmacogenomics data for a specified gene.
Description
This function queries the Open Targets GraphQL API to retrieve pharmacogenomics data for a specified gene.
Usage
pharmacogenomicsGeneQuery(ensgId)
Arguments
ensgId |
Character: ENSEMBL ID of the target gene (e.g., ENSG00000141510). |
Value
Returns a tibble containing pharmacogenomics data for the specified gene.
Examples
## Not run:
result <- pharmacogenomicsGeneQuery(ensgId = "ENSG00000141510")
## End(Not run)
Retrieve Pharmacogenomics data for a specified variant.
Description
This function queries the Open Targets GraphQL API to retrieve pharmacogenomics data for a specified variant.
Usage
pharmacogenomicsVariantQuery(variantId)
Arguments
variantId |
Character: ID of the target variant (e.g., "12_111446804_T_C"). |
Value
Returns a tibble containing pharmacogenomics data for the specified variant.
Examples
## Not run:
result <- pharmacogenomicsVariantQuery(variantId = "12_111446804_T_C")
## End(Not run)
Retrieve QTL Credible Sets data for a specified variant.
Description
This function queries the Open Targets GraphQL API to retrieve QTL credible sets data for a specified variant.
Usage
qtlCredibleSetsQuery(variantId, size = 500, index = 0)
Arguments
variantId |
Character: ID of the target variant (e.g., "19_10352442_G_C"). |
size |
Integer: Number of records to retrieve (default: 500). |
index |
Integer: Page index for pagination (default: 0). |
Value
Returns a tibble containing QTL credible sets data for the specified variant.
Examples
## Not run:
result <- qtlCredibleSetsQuery(variantId = "19_10352442_G_C", size = 500,
index = 0)
## End(Not run)
Retrieve Safety Liabilities data for a specified gene.
Description
This function queries the Open Targets GraphQL API to retrieve safety liabilities data for a specified gene.
Usage
safetyQuery(ensgId)
Arguments
ensgId |
Character: ENSEMBL ID of the target gene (e.g., ENSG00000141510). |
Value
Returns a tibble containing safety liabilities data for the specified gene.
Examples
## Not run:
result <- safetyQuery(ensgId = "ENSG00000141510")
## End(Not run)
Retrieve Shared Trait Studies data for specified diseases.
Description
This function queries the Open Targets GraphQL API to retrieve shared trait studies data for specified disease IDs.
Usage
sharedTraitStudiesQuery(diseaseIds, size = 500, index = 0)
Arguments
diseaseIds |
Character vector: IDs of the target diseases (e.g., c("EFO_0004587")). |
size |
Integer: Number of records to retrieve (default: 500). |
index |
Integer: Page index for pagination (default: 0). |
Value
Returns a tibble containing shared trait studies data for the specified diseases.
Examples
## Not run:
result <- sharedTraitStudiesQuery(diseaseIds = c("EFO_0004587"), size = 500,
index = 0)
## End(Not run)
Retrieve UniProt Variants data for a specified variant.
Description
This function queries the Open Targets GraphQL API to retrieve UniProt variants data for a specified variant.
Usage
uniProtVariantsQuery(variantId)
Arguments
variantId |
Character: ID of the target variant (e.g., "4_1804392_G_A"). |
Value
Returns a tibble containing UniProt variants data for the specified variant.
Examples
## Not run:
result <- uniProtVariantsQuery(variantId = "4_1804392_G_A")
## End(Not run)
Retrieve UniProt Literature data for a specified gene and disease.
Description
This function queries the Open Targets GraphQL API to retrieve UniProt literature evidence data for a specified gene and disease.
Usage
uniprotLiteratureQuery(ensemblId, efoId, size = 3500)
Arguments
ensemblId |
Character: ENSEMBL ID of the target gene (e.g., "ENSG00000130203"). |
efoId |
Character: EFO ID of the target disease (e.g., "MONDO_0004975"). |
size |
Integer: Number of records to retrieve (default: 3500). |
Value
Returns a tibble containing UniProt literature evidence data for the specified gene and disease.
Examples
## Not run:
result <- uniprotLiteratureQuery(ensemblId = "ENSG00000130203", efoId =
"MONDO_0004975", size = 3500)
## End(Not run)
Retrieve Variant Effect Predictor data for a specified variant.
Description
This function queries the Open Targets GraphQL API to retrieve variant effect predictor data for a specified variant.
Usage
variantEffectPredictorQuery(variantId)
Arguments
variantId |
Character: ID of the target variant (e.g., "4_1804392_G_A"). |
Value
Returns a tibble containing variant effect predictor data for the specified variant.
Examples
## Not run:
result <- variantEffectPredictorQuery(variantId = "4_1804392_G_A")
## End(Not run)
Retrieve Variant Effect data for a specified variant.
Description
This function queries the Open Targets GraphQL API to retrieve variant effect data for a specified variant.
Usage
variantEffectQuery(variantId)
Arguments
variantId |
Character: ID of the target variant (e.g., "4_1804392_G_A"). |
Value
Returns a tibble containing variant effect data for the specified variant.
Examples
## Not run:
result <- variantEffectQuery(variantId = "4_1804392_G_A")
## End(Not run)
Retrieve Variants data for a specified study locus.
Description
This function queries the Open Targets GraphQL API to retrieve variants data for a specified study locus.
Usage
variantsQuery(studyLocusId, size = 500, index = 0)
Arguments
studyLocusId |
Character: ID of the target study locus (e.g., "fa375739ca2a6b825ce5cc69d117e84b"). |
size |
Integer: Number of records to retrieve (default: 500). |
index |
Integer: Page index for pagination (default: 0). |
Value
Returns a tibble containing variants data for the specified study locus.
Examples
## Not run:
result <- variantsQuery(studyLocusId = "fa375739ca2a6b825ce5cc69d117e84b",
size = 500, index = 0)
## End(Not run)