The UCI gene expression cancer RNA-Seq ​dataset (https://archive.ics.uci.edu/ml/datasets/gene+expression+cancer+RNA-Seq) consists of 5 types of cancers (BRCA, KIRC, LUAD, COAD, PRAD) and gene expression levels of 20,531 genes for 801 samples. The dataset looks like this:

gene expression UCI dataset

How to query in The Cancer Genome Atlas (TCGA) to create an updated version of this dataset?

I've done the following query.

same query in visual form

Is the given query and the UCI dataset equivalent? If not, how to query properly?

P.S.: I'm a CS student; do not have any biology background.

  • $\begingroup$ Not quite what I was hoping for the query rather than the raw link ... Anyway I guess I wasn't clear $\endgroup$
    – M__
    Feb 18, 2022 at 16:57

1 Answer 1


To start with, your query above is selecting for breast cancer samples (TCGA-BRCA) only. The original dataset seems to contain 4 other cancer types.

The link you provided has a link to https://www.synapse.org/#!Synapse:syn4301332 and by looking at filenames of the 14 files there, I can tell that there are different types of data: miRNA expression, gene expression, methylation, ... And your query is limited to gene expression only. I am not even sure if you can use just a single query to fetch all of the 14 data files, you might need to use multiple queries.

If you are familiar with R, I would recommend TCGABiolinks, it will abstract away a lot of otherwise annoying manual steps. Apparently you can also use Python retrieve TCGA data:https://docs.gdc.cancer.gov/API/Users_Guide/Python_Examples/

  • $\begingroup$ I've added an image of the UCI dataset. I just need to perform a query that returns similar kind of gene expression level tabular data. As the query only returns TCGA-BRCA data only, will replicating the query for other four types create similar (biologically equivalent and meaningful) kind of dataset? $\endgroup$
    – fuad021
    Feb 19, 2022 at 17:25
  • 2
    $\begingroup$ Yes, you can change the cancer type and re-use the query but my main concern is the type of data that this query would return. Your query fetches count data via the HTSeq - Counts filter but in the posted image, there are floats instead of integers (gene counts), this can be normalized data. If I remember correctly, HTSeq-Counts should be returning actual counts, so integers. $\endgroup$
    – haci
    Feb 19, 2022 at 18:07

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.