Get Full Government Meeting Transcripts, Videos, & Alerts Forever!
Get email alerts on the Gene Expression Api topic
No spam. Unsubscribe anytime.
GDC gene-expression API: availability, selection and values endpoints explained
Summary
GDC demonstrated three gene expression API endpoints—availability, gene_selection and values—showing how to construct POST payloads with case UUIDs and Ensembl gene IDs, read top variably expressed genes (example TP53 and KRAS), and retrieve expression matrices in TSV format.
During the webinar Bill demonstrated the GDC’s gene expression API and the three endpoints that underpin the portal clustering tool: availability, gene_selection and values.
He explained how to build cURL POST requests and emphasized payload conventions: case identifiers must be UUIDs (exportable from the cohort builder) and gene identifiers must be Ensembl IDs (gene symbols are not accepted for the API). Bill warned that curly quotation marks (often introduced by Word/PowerPoint) break command lines and advised using straight quotes. He also recommended the command-line JSON formatter jq for more readable terminal output.
Availability endpoint: this endpoint returns which supplied cases and genes have expression data; Bill showed an availability response that includes a details key for cases and genes with true/false markers and a summary count (e.g., “10 genes with gene expression, 0 without”).
Gene selection endpoint: users can request the top N most variably expressed genes within a given set of cases (or across all protein-coding genes by specifying gene_type=protein_coding). In the demo Bill requested the top 2 variably expressed genes for the example cohort and reported the results as TP53 and KRAS with associated log2 UQ FPKM medians and standard deviations.
Values endpoint: this endpoint returns expression values for a specified list of cases and genes in TSV format. Bill noted supported units include UQ FPKM (used by the portal) and median-center log2 UQ FPKM; output TSVs have gene IDs in the first column and case UUIDs as column headers. Because UUID headers are long, Bill recommended parsing TSV output with Python, Excel, or Google Sheets rather than reading it raw in a terminal.
Practical tips from the demo: export cohort UUIDs through the cohort builder for reliable case identifiers; export gene sets as TSVs to extract Ensembl IDs when needed; use jq to prettify JSON output for availability/selection responses; and double-check quotation characters when copying example commands from slides.
Documentation and example payloads are available on the GDC documentation site (docs.gdc.cancer.gov) and the GDC Apps pages; Bill closed the API demo with a reminder that the clustering tool itself is implemented on top of these endpoints.
