AWS HPC Blog

Helping bioinformaticians transition to running workloads on AWS

This post was contributed by Swaine Chen, Amazon Scholar, APJ Health; Austin Cherian, Sr. Product Manager, HPC; Sarah Geiger, Postdoctoral fellow, National University of Singapore; and Suma Tiruvayipati, Postdoctoral fellow, National University of Singapore

Introduction

Genomics and bioinformatics are transforming all aspects of biomedical research, especially through increases in scale and analysis complexity. Key drivers of these changes are:

  1. Growing data sets from increased output and lowered cost of DNA sequencers and other high-throughput technologies
  2. Improved algorithms enabled by advances in computing, particularly in the cloud
  3. Better and larger online databases being built by research communities
  4. Accelerating expectations in grants and manuscripts for innovating at scale and sophistication

These are exciting times! However, researchers also potentially face challenges with emerging technology, and support from institutions may vary with available resources and expertise. Therefore, while individual breakthrough results from new technology applications might grab headlines, raising the overall research community’s expertise on emerging technologies is also critical and potentially more impactful.

Identifying common challenges

There are numerous resources available to help bioinformaticians learn and work. These span the range from online forums and courses to books and published articles in the scientific literature. Many times, local user groups, the local IT or scientific computing team, or even “asking that person in the other lab” is where support happens. There are also web applications, recipes and code snippets, frameworks and workflow managers, and integrated development environments (or IDE-like solutions like Galaxy or Jupyter notebooks). There are containers and code repositories and online data stores and cloud resources. This rich ecosystem can be daunting for a new user.

We saw a common set of problems among research bioinformaticians, especially with beginning researchers, across four major categories:

  1. Acquiring computing resources. For example, installing a Linux virtual machine on a laptop or getting an account on a university/department cluster. Researchers face many options and technical tradeoffs that beginning users lack context within which to properly decide.
  2. Locating or installing software. Accessing bioinformatics software usually requires some minimal understanding of system administration and package managers or of operating in shared clusters with managed software, such as loading environment modules.
  3. Accessing data. Moving data between different locations often requires basic knowledge of networking, client-server architecture, and authentication. More seamless configurations, such as common shared drives or host-guest integration, may require even deeper knowledge to set up initially.
  4. Analyzing data. Finally, after all this, you can start analyzing data! This is where the diversity of tools and scale of data can generate additional challenges.

Improving the process

With cloud computing, we believe several of these issues can be made easier for a large proportion of budding bioinformaticians. Specifically, cloud computing on AWS provides services that are ideal for completely solving the first challenge of acquiring computing resources. AWS offers virtual servers with Amazon Elastic Compute Cloud (Amazon EC2) and storage with Amazon Simple Storage Service (Amazon S3). Amazon Machine Images (AMIs), the AWS Marketplace, and container registries, such as Amazon Elastic Container Registry (Amazon ECR), can help with the second challenge of getting installed software.  When public data is required (the third challenge above), the Registry of Open Data on AWS can facilitate access to a growing collection of data sets.

The first and second challenges are undifferentiated heavy lifting, and researchers want to quickly move to the third and fourth challenges that are more specific to bioinformatics work. Therefore, to improve the process, we developed a hands-on workshop. The workshop mixes short lectures on new vocabulary, concepts, and considerations with a guided practical tutorial.  The tutorial guides new users to use AWS, provision Amazon EC2 instances, choose from provided AMIs with a curated set of pre-installed software, and move data between their own local laptop or computer and the EC2 instance. The tutorial guides the audience in understanding how to stitch these various steps together to run real world bioinformatics analyses taken from bacterial genomics, long read sequencing, and single cell expression profiling. The tutorial itself is implemented as a website with screenshots and explanations. Users can work at their own pace and refer back to the website in the future. With AWS, you pay for only the services that you use, so running the workshop on one’s own has an estimated total cost of less than $5 USD. If users are eligible, much of it can be done using resources available in the AWS Free Tier.

Attending a workshop

We delivered this “Introduction to Genomics on the Cloud” workshop eight times over the past 18 months. Through online-only events, we reached over 200 participants in 10 countries. The workshops focused on explaining cloud computing concepts and facilitating the hands-on tutorial. We avoided “magic commands” that are just copied and pasted, instead trying to illustrate cloud and analysis concepts. We also covered cost and security concepts and tools so that participants can confidently continue experimenting on their cloud journey after the workshop.

Feedback has been quite positive, with attendees both satisfied and willing to recommend the training to others. We thus continue to have strong interest in additional sessions, validating that we have identified and addressed issues that resonate with a segment of the research bioinformatics community. One previous participant noted:

I wanted to get a feel of AWS…before starting out on my own. I did not want to set up things without understanding and get shocked by a huge bill at the end of the month. This workshop gave me exactly what I was looking for. Now I feel confident to set up my own instances and buckets, and know exactly what are the things to keep in mind.

We continue to plan additional workshops. Please register online if you are interested in attending a future session!

Conclusion

Bioinformaticians are as diverse as their research interests. Although we tried to address common concerns for a subset of bioinformaticians, others experience different challenges. In our next post, we dive a bit deeper into another part of this training, our AMI, which provides a good general base system for beginning bioinformaticians. In addition, we discuss how we constructed our AMI to help researchers extend it. You can also learn more about our workshop in the HPC Tech Shorts video series.

TAGS: ,
Swaine Chen

Swaine Chen

Swaine Chen is an Amazon Scholar in Healthcare at AWS, an Associate Professor of Medicine at the National University of Singapore, and a Group Leader in Bacterial Genomics at the Genome Institute of Singapore. His academic research work combines his training in medicine, biology, chemistry, and mathematics to develop genomic approaches for understanding how and why bacteria are able to cause infections in humans. His lab has several experimental biologists who have fully converted to computational research, and he continues to have a passion for expanding the field of bioinformatics through new recruits.

Austin Cherian

Austin Cherian

Austin is a Senior Product Manager-Technical for High Performance Computing at AWS. Previously, he was a Snr Developer Advocate for HPC & Batch, based in Singapore. He's responsible for ensuring AWS ParallelCluster grows to ensure a smooth journey for customers deploying their HPC workloads on AWS. Prior to AWS, Austin was the Head of Intel’s HPC & AI business for India where he led the team that helped customers with a path to High Performance Computing on Intel architectures.

Sarah Geiger

Sarah Geiger

Sarah Geiger, PhD, is a postdoctoral research fellow in the lab of Swaine Chen at the National University of Singapore and the Genome Institute of Singapore. She has been developing genomic methods to understand outbreaks of infectious diseases, all using AWS Services. She is one of the creators and teachers for the GIS x AWS Bioinformatics Cloud Workshop.

Suma Tiruvayipati

Suma Tiruvayipati

Suma Tiruvayipati, PhD, is a postdoctoral research fellow in the lab of Swaine Chen at the National University of Singapore and the Genome Institute of Singapore. She has been developing genomic methods to improve processing methods relevant for molecular epidemiology and outbreak analyses, all using AWS Services. She is one of the creators and teachers for the GIS x AWS Bioinformatics Cloud Workshop.