Tissue-Level Expression: GTEx
Estimated time: 25 minutes
Use GTEx to compare expression of the 25 genes in atrial appendage and left ventricle.
Tissue-level expression
The study narrowed millions of variants to 54 observations across 25 genes. We will use GTEx to compare median expression of these genes in two reference heart tissues.
What GTEx contains and why we use it
GTEx stands for Genotype-Tissue Expression. This resource contains gene-expression data from human tissue donors.
GTEx represents a general reference donor cohort. We will compare atrial appendage and left ventricle, the two heart tissues represented in GTEx.
Expression is reported as median transcripts per million, or TPM. TPM summarizes abundance while accounting for transcript length and sequencing depth. The median gives one reference value across the GTEx samples for a tissue.
Learn more about the resource and its endpoints in the GTEx Portal API documentation.
How to read GTEx values
Treat the median TPM values as a descriptive comparison between reference tissues. Because GTEx measures bulk tissue, each value combines signals from all cell types present in those samples.
GTEx adds an atrial-appendage and left-ventricle reference value for each gene. A higher median in one tissue can guide tissue selection for follow-up.
Querying the GTEx API
The live request begins by matching the paper’s 25 gene symbols to GENCODE v39 IDs. Once those identifiers are resolved, a second request retrieves GTEx v10 median expression for the two heart tissues. Specifying v10 keeps the live request aligned with the dated teaching response.
Prepare the gene list and API helper
Load the published variants, select the 25 unique gene symbols, and import the GTEx wrapper.
gene_symbols contains the 25 unique gene symbols in alphabetical order. variants contains all 54 published rows.
Request heart-tissue expression
Request median expression for the 25 genes in atrial appendage and left ventricle. In the returned table, look for the gene and tissue identifiers, the median TPM value, and the retrieval date.
One row represents one gene in one tissue. median_tpm records median transcript abundance.
The gencode_id, dataset_id, and retrieved_date fields record the source and retrieval, while unit confirms TPM.
Check the returned data
Before comparing tissues, make sure the response is complete. This request should return 25 genes, 50 gene-tissue rows, two tissues, and TPM units.
If a gene or tissue row is missing, this summary exposes the problem before we reshape or interpret the expression values.
Values are requested directly from the GTEx v10 API. API availability can change, so the dated v10 teaching response provides a backup.
Request sequence
The fetch_gtex_context() wrapper resolves gene symbols, requests expression, and returns a tidy table.
| Step | Endpoint | Important parameters |
|---|---|---|
| Resolve symbols | /api/v2/reference/gene |
Gene symbols, GENCODE v39, and GRCh38 |
| Request expression | /api/v2/expression/medianGeneExpression |
GENCODE IDs, GTEx v10, atrial appendage, and left ventricle |
The complete implementation and endpoint constants are maintained in api_helpers.py.
Compare heart tissues
Reshape the response into a gene-by-tissue table, add the number of published rows per gene, and plot a readable subset.
Summarize genes and tissues
Reshape the GTEx values so each gene has one column per tissue. Then add the number of published rows connected to each gene for reference.
Each gene now has one row. The tissue columns contain median TPM, while variant_rows counts appearances in the published table.
In the dated teaching data, DES and MYH7 have the highest left-ventricle medians.
Plot genes with the highest expression
Plot the ten genes with the highest median left-ventricle expression. This smaller view keeps labels readable. The complete table above includes all 25 genes.
In the dated teaching data, DES and MYH7 have higher median expression in left ventricle than atrial appendage, while ACTC1 is higher in atrial appendage.
These bulk-tissue results lead to the cell-type question in the HuBMAP lesson.
Compare the atrial ranking
Rank the same genes by atrial-appendage expression and compare the result with top_heart_genes.
Use the Heart - Atrial Appendage column.
# Rank genes by atrial-appendage expression.
top_atrial_genes = heart_expression.nlargest(
10,
"Heart - Atrial Appendage",
)
top_atrial_genes.head()ACTC1 ranks first in atrial appendage, while DES ranks first in left ventricle. LMNA appears in the atrial top 10, while MYL3 appears only in the left-ventricle top 10.
These differences show that the tissue selected for comparison affects the gene ranking.
Check your understanding
What does median_tpm represent in this lesson?
Correct. Each value summarizes gene-level expression across the GTEx samples represented for one tissue.
The values come from the GTEx reference resource. The heart-failure study cohort supplied the candidate variants.
The separate variant_rows column records how often a gene occurs in the published candidate genetic variant table.
What does GTEx add to the published variant list?
Correct. GTEx describes gene expression in the selected heart tissues.
The paper supplies the classification for each published variant row.
Key points
- The GTEx v10 response contains median TPM for all 25 genes in two bulk heart tissues.
- DES and MYH7 have the highest left-ventricle median TPM in the dated data.
- Ranking by atrial appendage instead of left ventricle changes the top-10 gene set.
Next: GTEx summarizes bulk heart-tissue samples, which contain several cell types. Use HuBMAP to ask whether the same genes have indexed expression in ventricular cardiac myocytes.