Download PCAWG Data

Data Repositories

Data produced by the PCAWG study has been deposited into multiple long-term storage systems. This table gives you an overview where data for different types can be downloaded.

Repository Storage type Data types Projects
ICGC Data Portal (Portal) Web Accessible File Downstream analysis results, cell line benchmarking, QC, reference genome etc All Non-US projects
Cancer Genome Collaboratory (Collab) Data Object Aligned WGS, RNA-Seq, variant calls All Non-US projects
Amazon Web Service (AWS) Data Object Aligned WGS, RNA-Seq, variant calls Subset of Non-US projects
European Genome-phenome Archive (EGA) Data Archive Aligned WGS, RNA-Seq All Non-US projects
Protected Data Cloud (PDC) Data Object Aligned WGS, RNA-Seq, variant calls, downstream analysis results All US projects (TCGA projects)

NOTE: Subset of Non-US projects with policy permits data to be hosted at AWS can be found from here.

Apply for Data Access

Most of the PCAWG data is controlled access that is subject to data usage agreement. Please follow instructions at ICGC DACO to apply for access to controlled data of all non-US projects. For US TCGA portion of the PCAWG data, apply for access through dbGaP.

Browse / Search for PCAWG Data

Although PCAWG results are physically stored in multiple repositories, the ICGC Data Portal provides a one-stop-shop entry point to browse / search for all PCAWG data.

Browse PCAWG Data

In the portal's Data Release section, there is a dedicated PCAWG directory where you can browse study-wide datasets (such as cell line benchmarking, QC, reference sequences) and downstream analysis results (usually associated with PCAWG publications).

The data portal presents a simple interface allowing users to navigate through the entire PCAWG data directory. Besides the data files, most of the sub-directories contain a README file that gives a summary of data content, and provides additional information about how to access related data in external systems. For example, the README in germline_variations describes how the germline callset was generated, and where ICGC portion and TCGA portion of the results can be downloaded.

Search for PCAWG data

A faceted search interface at the ICGC portal offers an easy way to interactively search for PCAWG data of your interest.

This particular query set a filter on Study = 'PCAWG', it returns all PCAWG data objects. Results are listed in a table with File ID, Donor ID, Repository etc important information. You can add more filters as needed, for examples, adding RNA-Seq in Experimental Strategy, will further narrow down the resulting data objects to PCAWG RNA-Seq. Here is the new, resulting query. If you prefer to download data from certain repositories, you can select the desired repositories as needed.

After identifying files of interest, click on the Download Files button which will bring up a dialog window displaying a summary of file counts and total size broken down by repository. As many files are distributed in multiple locations, you may want to prioritize on certain repositories. You may enable Remove duplicate files as well, it will remove files that exist in a repository that was already presented. Finally, you can click on Download Manifest which will download a tarball of manifest files, one for each chosen repository.

Manifest file format differs from one repository to the other, and thus each manifest must be used with the corresponding client tool to actually download the data. The following section will cover how to perform data download from different repositories using manifest files.

Download Data

Download from ICGC Data Portal

As mentioned earlier, data files hosted in ICGC Data Portal can be browsed in a web browser, downloading specific file of interest is a matter of a mouse click.

Open access data hosted under the PCAWG directory and its subdirectories can be downloaded without logging in. However, you must login to the Data Portal before you are able to download or even see the controlled access data files. Click on the Login button on the top right corner, and follow steps. Once successfully logged in, you should be able to see two green icons (beside your account name) like shown below on the top row of the portal. This assumes that you have already been approved by ICGC DACO for accessing controlled data.

Once logged in, you will be able to browse the PCAWG directory as usual with controlled access data files visible and downloadable.

Retrieve Access Tokens from ICGC Data Portal

When logged in you should be able to use the Token Manager (as shown above) to create access token for yourself to be used to download data from Collaboratory and AWS (see more details on how in the next section). Tokens expire in a year, you can delete a token before expiry at anytime.

Download from Collaboratory or AWS

Score Client is used to download from Collaboratory or AWS with a TSV manifest file containing necessary information about data objects to be downloaded.

The latest version of Score Client can be downloaded from here. If you are on a Linux system and already have Java 8 installed, you can have Score Client ready to use with the following commands:

wget -O score-client.tar.gz[RELEASE]/score-client-[RELEASE]-dist.tar.gz

tar xvzf score-client.tar.gz

echo export PATH=$(pwd)/score-client-x.x.x/bin:$PATH >> ~/.bashrc

source ~/.bashrc

# you will then be able to use it by running the following command anywhere is the system

Before you can actually download controlled access data, you will need to add the ICGC access token in the following file: score-client-x.x.x/conf/ In the file, you should see a line like shown below. Uncomment it, add your access token, and save the file.

# accessToken=your_collab_access_token

Assuming that you downloaded a manifest file (manifest.collaboratory.1525977569066.tsv) from ICGC data portal in the previous step, the manifest will contain files from Collaboratory. The following command will download these files to the current directory:

score-client --profile collab download --manifest manifest.collaboratory.1525977569066.tsv --output-dir .

NOTE: Downloading data objects hosted in Collaboratory is no longer required to be performed in a Collaboratory compute instance. When downloading PCAWG data from AWS, the download client must run in the same environment as the object storage system. That means to download from AWS (N. Virginia), the Score Client must run on an AWS VM in the N. Virginia availability zone.

Download from EGA

After getting your ICGC DACO approval, you will need to get in touch with EGA to have your EGA account set up. You will then be able to download any EGA Datasets that are under the control of the ICGC DACO.

For EGA data download, the manifest file you download from ICGC Data Portal is actually a shell script with a few variables need to be replaced by actual values, such as, username, password, output_dir etc. The script utilizes EGA Java download client which is documented with more details here. You may modify the manifest script to suit your own need. For example, if download involves too many files it could be inefficient to request all of them at once. You may change the script so that it requests one file at a time, finish downloading and decrypting one file, then move on to the next file.

Download from PDC

Download manifest file from ICGC Portal

As described in the Search for PCAWG data section, once you satisfied with the search result of TCGA data files, click on the "Download Manifest" button as illustrated below to retrieve the manifest tarball (named as manifest.*.tar.gz). Unpack the tarbal, you should get a file named as manifest.pdc.*.sh, eg,

Convert ICGC manifest file to PDC's Gen3 manifest file

We need to convert the mainfest file to PDC's Gen3 manifest file before downloading the actual data files. A Python script ( is needed to perform the conversion, the script can be downloaded with the following command:


You need to have Python 3 and required libraries (such as numpy and pandas) installed. Once installed, you can run the script to get Gen3 manifest file. Remember to replace the ICGC manifest with your own file name.

python --manifest

This will produce a Gen3 manifest file named as, which contains information needed to download the acctual data from PDC using gen3-client tool.

Install Gen3-client

Run the following commands to install gen3-client if you are using macOS:

mkdir -p ~/.gen3
echo "" >> ~/.bashrc
echo "export PATH=\$PATH:~/.gen3" >> ~/.bashrc
curl | grep browser_download_url.*osx |  cut -d '"' -f 4 | wget -qi -
mv gen3-client ~/.gen3
source ~/.bashrc

With that you should be able to run gen3-client command from your console and see the usage message.

For installing gen3-client on other OS, please follow instructions here.

Get gen3-client API key and configure your profile

Now you need to create gen3-client API key from after authentication via NIH eRA commons. To do that goto login page, and click on "Login with NIH" button. After authenticated successfully, please goto to create the API key. On the popup dialog click on "Download json" to retrive API key, as shown below:

The API key will be saved as credentials.json. You can then use it to configure a profile, let's name the profile icgc:

gen3-client configure  --profile=icgc --cred=credentials.json --apiendpoint=

Upon success, you should see a message: Profile 'icgc' has been configured successfully.

Download data using gen3-client download-multiple command

With icgc profile configured, you can download the PCAWG data using the gen3 manifest prepared earlier as follow:

gen3-client download-multiple --profile=icgc --no-prompt

Analyze PCAWG Data on Collaboratory

Leverage compute and data colocation

As noted above, a large portion of the PCAWG data is hosted at the Cancer Genome Collaboratory - an OpenStack based academic cloud computing platform. This makes Collaboratory an ideal choice to perform your own analyses on PCAWG data, following a commonly practised move compute to data approach. In essence, there is no need to download large amount of data into your own computing facilities and manage them, instead you launch virtual machines (VMs) and run your code on PCAWG data provisioned from within the Collaboratory environment which is much more efficient and cost effective.

Request a Collabotory account

You can find out more about services provided by Collaboratory at and available cloud resources at

To request a Collaboratory account please visit: