Introducing Overture

Dec. 8, 2017

Today we are introducing Overture – our collection of open source, composable and extendable components for data science in the cloud. Although only introduced today, its inception started as a concept months ago, as we were receiving requests from various organizations for guidance on how to run some of the ICGC components in their own environment, with their own data or with a mirror of ICGC data.

As internal discussions progressed on documenting our ICGC tools in the perspective of them being used outside of the team, we started to better structure our open-source repositories and created an ICGC Github organization to separate tools specific to our data portal from reusable tools. We soon realized this was not enough, that we were, with all our projects (ICGC-DCC, Collaboratory, NCI-Genomic Data Commons, Kids First DRC, and more to come), in a privileged position to understand, design, implement and deploy components spanning the entire field of genomic data commons. And more than that, some of those components were already built and mature. Using years of experience operating the ICGC Data Coordination Centre, we learnt from our mistakes, optimized implementations and overall built very robust and scalable components for our field.

Our projects

To better understand why we are in a privileged situation with Overture, it’s worth detailing the scope of our main projects:

  • On Cancer Gernome Collaboratory our team is building a cloud environment from scratch. This includes purchasing hardware, designing and operating the infrastructure (running OpenStack and Ceph), operating a mirror of the ICGC Dataset with its related toolset and researching possible ways of implementing a cost recovery model. To do all of this, we are using off-the-shelf tools (OpenStack, Ceph, Zabbix, …), but also building our own when such tools don’t exist.
Collaboratory Racks. Compute at the top, storage at the bottom.
  • On ICGC-DCC, we are building tools to support the entire data portal lifecycle, receiving data, processing it through a complex ETL, and making the resulting dataset available in a Data Portal. Started more than 5 years ago, it progressively went through different implementations, technological evolutions (ETL rewrite, multiple ES upgrades…), allowing the team to gain experience and better understand long term viable solutions for our field.
Architecture diagram for Song and Score (Metadata and file storage components)
  • On Genomic Data Commons, we are building the front ends and part of the (GraphQL) API of this large scale project. The GDC allowed us to start a UI from scratch, improving on lessons learnt from the ICGC. The portal and API went through many iterations, migrating from AngularJS to React and from a REST API to GraphQL. Starting from scratch provided us with flexibility for innovation, something that could have been slightly more challenging with the ICGC due to the size of the codebase. GDC was also the opportunity to start OncoJS to share efforts on visualization components between our projects (a first step towards Overture).
Oncogrid, an OncoJS component shared between GDC and ICGC
  • On Kids First, we are building upon our experience with the 3 above-mentioned projects, but are faced with a new and exciting challenge, catering not only for researchers and field experts but also for patients and their families. This involves the creation of new components, with features more traditional of a genomic data portal (file repository, various entity pages…) cohabiting with features more common on social media platforms.
A knowledge transfer on ETL (yes, lab coats were worn just for the picture)

Share components

As you might guess from the above, we try as much as possible to share expertise and components between our team and projects. But some of them were (and are still) opinionated to the way our team operate and to the specificities of our projects.

That’s where Overture start to play a role, detaching our components from specific projects and embedding them under the “Overture” umbrella will encourage us to implement them in a reusable fashion. Our projects will become users of Overture at a level similar (although slightly privileged, as you can guess) than any external user.

Four categories of components

Detailed on Overture’s website, we divided the components in four categories:

  • Operate: Components focused on the operations of a cloud environment and its tools. Currently focused on infrastructure, it will extend to more, such as admin UIs.
  • Transfer & Store: Securely transferring and storing data can be complex, tools in this category will be dedicated to topics such as authentication, authorization, download, upload, transfer management.
  • Do Science: Once the environment is in operation, tools in this section will be focused on actually running analysis or workflows on the data. From running a Jupyter notebook on ICGC data to distributed computing on genomic files.
  • Share: The data is there, it has been analyzed, now is the time to make it available to the research community. This category will focus on the dissemination of content, whether it’s in the form of datasets or modern UI and visualization tools.

Where do we stand today ?

We are still at the beginning of the adventure. Our most recent components (ego, song, persona) were implemented in a reusable fashion and would be from the get-go aligned with Overture goals. We will progressively iterate, transition and document existing components (such as SCORE, our storage/data transfer solution) into Overture. Ensuring all of the components retain a level of cohesion in their implementation and documentation.

Overture organization on GitHub

We released an initial version of the website, listing the components already in Overture or that will soon be ported over. The website is still in its early stages, but should give a good overall picture on what Overture is. Each of the component will be further detailed and have their own page on the website, with instructions to get started quickly.

We are still early in the process, but today is a good time for an initial announcement.

Next ?

Openness is, and will remain, at the core of Overture. we will progressively transition our project management tools (i.e. Jira) to a public platform (i.e. GitHub issue) to give everyone a vision on our implementation roadmap and priorities. We also welcome external contributions to enhance the tools, get external input and overall make the components better.

Some components are already part of Overture, more will come shortly.

You can learn more on our website: and of course on Github:

PS: GDC is a joint effort with the University of Chicago. Kids First is a joint effort with the Children Hospital of Philadelphia, University of Chicago, CNHS, OHSU and SevenBridges

Francois Gerthoffert, Project Manager
Passionate project manager, diehard curious and solutions finder. Enjoy learning and experiencing new things, from rollerblading to Niagara falls to making his own furniture. Did I hear someone talking about cheese ?