## Changes in version 2.14.0 ### Bug fixes and minor improvements * Remove error when `studyId` build success is unknown (@vlaufer, #69) * Parse tokens with different formats, e.g. genie token (@ZWael, #70) * Clean up duplicate rows created in the `sampleMap` generated from the data * Fixed warning trigger when empty `molecularProfileId` datasets are found ## Changes in version 2.12.0 ### Bug fixes and minor improvements * Update `successrate` thresholds and fix long tests ## Changes in version 2.10.0 ### New features * Add `cgdsr` to `cBioPortalData` migration vignette (@kmezhoud, #54) * Unmapped experiments are now added to the metadata in `cBioDataPack` * Set default `api. = api/v2/api-docs` in `cBioPortal` to access the API protocol's new location * The `fetchData` developer function added to handle both `molecularData` and `mutationData` requests as deduced from the `molecularProfileIds` vector ### Bug fixes and minor improvements * Check for valid `studyId`s with `getStudies` in `cBioDataPack` * `ask` argument correctly passed down to caching mechanism in `cBioDataPack` * `check_build` option available in `cBioDataPack` particularly for new studies that have not been checked against. ## Changes in version 2.8.0 ### New features * Auth `token` string or file can now be included in the `cBioPortal` function. * The `check_build` argument can be set to `FALSE` for alternative APIs, e.g., KidsFirst, when using `cBioPortalData` * `queryGeneTable` translates gene IDs ('hugoGeneSymbols' <> 'entrezGeneIds') via the API service * `getDataByGenes` supersedes `getDataByGenePanel` * `getStudies()` replaces `data('studiesTable')` to discover study IDs ### Bug fixes and minor improvements * Fixed issue where the `by` argument was not passed to `getDataByGenes` in internal calls * Add names to metadata elements that originate from GISTIC datasets. ## Changes in version 2.6.0 ### New features * A study's build status can be obtained from `getStudies()`, which has replaced `data('studiesTable')`. * Partial loading of data files supported. A warning is emitted when a data file is not able to be loaded in `cBioDataPack`. * `cBioPortalData` checks the `data(studiesTable)` to verify that study datasets are building, otherwise provide a message in interactive sessions. ## Changes in version 2.4.0 ### New features * Vignettes include additional information (#38, @lwaldron) * `getDataByGenePanel` deprecated for `getDataByGenes` which handles input of both gene panels and genes * `cBioPortalData` now allows for gene inputs as either Entrez IDs or Hugo symbols (#24, @jucor) and `sampleIds` input * When `gene` inputs are provided, the `by` argument has to agree with the type of genes provided (either be `entrezGeneId` or `hugoGeneSymbol`). ### Bug fixes and minor improvements * Fixed an issue where the labels in the `metadata` from `cBioDataPack` were missing ('LICENSE' and 'Fusion'; #37) * `loadStudy` allows `cleanup=TRUE` for removing files after `untar`-ing * Published article now available with `citation("cBioPortalData")` ## Changes in version 2.2.0 ### New features * `studiesTable` includes additional columns `pack_build` and `api_build` to indicate to the user which datasets have been successfully built as `MultiAssayExperiment` objects. Users will be notified when a dataset, reported as not building, is requested from the `cBioDataPack` function. * Add `sampleIds` argument to `getDataByGenePanel` as part of cache re-work * Allow more flexibility in the hostname when accessing the API with `cBioPortal` (@inodb, #16) * `cBioDataPack` downloads from a more robust repository (AWS S3; @inodb, #22) * `removePackCache` and `removeDataCache` now remove data from the user's cache based on inputs to respective functions (`cBioDataPack` and `cBioPortalData`) ### Bug fixes and minor improvements * Attempt to merge additional clinical data files from tarballs in `cBioDataPack`. * Switch to using `read.delim` instead of `read_tsv` internally to avoid assigning `NA` to chromosome column * Use 'PATIENT_ID' when available to determine if experiment data is provided in the tarball files. * Add tests using `testthat` * Update and include percentages of studies successfully imported using `cBioDataPack` and `cBioPortalData` in the documentation * Fix read-in when identifiers are numeric instead of character (@jucor, #27) * Include pagination parameters in `geneTable` function (@xinwei-sher, #29) ## Changes in version 2.0.0 ### New features * Bioconductor release! * Updated the `README.md` file from R Markdown file. * Uses the latest version of `rapiclient` on CRAN * Prepare package for Bioconductor submission * Include protein metadata as a `RaggedExperiment` from mutation molecular profiles (TCGA only) ### Bug fixes and minor improvements * API authentication option removed and not needed ## Changes in version 1.0.1 ### New features * Package supports nearly all study identifiers based on recent tests * Only a handful of study identifiers are unsuccessful (create an issue to prioritize). ### Bug fixes and minor improvements * Make better use of the API return values to craft the sample map for `MultiAssayExperiment` creation * Additional data included in the metadata slot of the `MultiAssayExperiment` object. Future revisions will include this data as `rowData`. * Change vignette titles for build ## Changes in version 0.1.0 ### New features * `cBioDataPack` allows users to download packaged data objects from download.cbioportal.org/ * Data packs are cached using `BiocFileCache` to avoid re-downloading * `cBioPortalData` lets users query the cbioportal.org API and retrieve slices of data according to gene, molecular profile identifiers, etc. * Queries through `cBioPortalData` use a caching mechanism to avoid repeat downloads of data and improve load times * Both functions return a `MultiAssayExperiment` as the primary data representation * Only a number of study datasets are currently possible to load. Issues can arise with mismatched or munged identifiers * The cBioPortal API representation is handled by the `AnVIL` package which makes use of `rapiclient` to provide an automatic R interface to the API ### Bug fixes and minor improvements * Data pack downloads use an alternative method for download when a download fails