Accelerating Drug Discovery From One Year to Seven Hours

Accelerating Drug Discovery From One Year to Seven Hours – OMass Therapeutics

Learn how YellowDog helped OMass Therapeutics accelerate its drug discovery process from 1 year to 7 hours with an on-demand, 378,432 core supercomputer in the cloud.

Title image "Accelerating drug discovery from one year to seven hours"

OMass Therapeutics is an Oxford biotechnology company spun out of Oxford University harnessing native mass spectrometry and other biophysical technologies to drive drug discovery in high definition. The company’s suite of proprietary technologies delivers superior resolution to traditional pharmacology assays enabling a clearer, more detailed picture of how a putative ligand engages a protein target, assessing binding and function in parallel. The company is building a pipeline of small molecule therapeutics to bring life-changing benefits to patients suffering from immunology and genetic disorders. OMass is backed by Syncona Ltd, Oxford Sciences Innovation and Oxford University.

Learn how YellowDog helped accelerate OMass Therapeutics’ virtual screening process from one year to seven hours, with its powerful multi-cloud workload management solution.

There is great potential for the cloud to and the process of drug discovery

The Potential of Using the Cloud for Drug Discovery

For today’s drug discovery processes, as with many of the life sciences, data is key. The size of a dataset and the number and quality of the analyses has a powerful influence on the potential success and efficacy of a given drug development programme. The larger the population of data points, the faster you can carry out the analyses, and the greater the probability of success when identifying new chemical matter as targets of interest.

There is great potential for the cloud to aid the process of drug discovery as it offers companies the ability to dynamically scale their research and analyse more data than they would with their traditional on-premise infrastructure. Considering the commercial and societal value of new drugs and therapeutics, any acceleration to the process can result in substantial benefits.

Overcoming the Barriers to Cloud Adoption for Drug Discovery

Although the cloud can help biotechnology companies, such as OMass, a fear of uncontrolled costs can often hold companies back: with unpredictable variable costs, potential costs for failed tests, and traditional on-premise analysis applications licensed in such a way that doesn’t fit with the way the cloud works.

Then there are the technical challenges of securely connecting to, and provisioning computing power in the cloud. Finding the right compute and storage resources that meet the needs of the dynamic, compute-intensive workload can be difficult. With over 25,000 individual cloud instances available today across the main public cloud providers, it’s not always easy to find the best source of compute – whether for price, performance or compliance requirements.

Image of laboratory with text "With YellowDog OMass were able to dock 650 times more virtual compounds"

Processing Molecules in Record Time

Harvard University’s REAL database contains over 1.4 billion commercially available on-demand molecules. This database can be used by companies to find potential hit compounds for the development of new drugs and therapeutics.

OMass saw an opportunity to accelerate and expand its virtual screening capabilities by utilising this large compound database and YellowDog’s multi-cloud workload management solution. With access to an on-demand supercomputer, OMass was able to dock 650 times more virtual compounds, and significantly enrich the top 10% of compounds screened. This top-level analysis enabled OMass to focus its attention on the most likely active molecules for its specific drug target.

Based on an initial set of filters, OMass presented YellowDog with a refined list of 337 million molecules for virtual screening. To achieve this scale, YellowDog delivered:

Fast provisioning and intelligent orchestration of 4,000 AWS Spot instances, a total of 378,432 cores, across four geographic regions.
Task distribution and re-distribution when Spot instances were reclaimed by AWS to ensure job delivery.
Identification of the Best Source of Compute for the workload; based on cost, processing power and availability.
Integration with AutoDock Vina, an Open Source application for molecular docking.
Secure storage and data retrieval using YellowDog Object Store.
Achieving a 95% vCPU utilisation across all instances.

A graph to show the points at which Spot instances were reclaimed by AWS and then reprovisioned by YellowDog’s intelligent Scheduler Service — Figure 1 – The above graph shows the points at which Spot instances were reclaimed by AWS and then reprovisioned by YellowDog’s intelligent Scheduler Service. There is very little effect on productivity with the yellow line, showing running instances, closely mirroring the blue line, which shows the number of instances requested.

Significantly Accelerating Virtual Screening From One Year to Seven Hours

The outcomes for OMass are outstanding:

The 337 million compounds were analysed and screened in a total of 7 hours. Using OMass’ existing setup, the same number of analyses would have taken 1 year.
By finding the Best Source of Compute, YellowDog delivers a cost saving of 56.9% on the cloud compute costs for the workload.
YellowDog is fast: provisioning the 4,000 cloud instances in AWS and getting them to process the workload took less than 5 minutes. Using a standard scheduler or cloud provisioner would typically take 20 hours.
All tasks and queues were populated in 3 minutes.

Figure 2 – A graph to show the time it took to provision 4,000 instances using AWS. Within 5 minutes all instances were running.

Bringing Life-Changing Benefits to Patients

The radical order change in the number of molecular docking analyses that are able to be carried out at this early stage in the drug discovery process, coupled with the speed and cost effectiveness of the YellowDog Platform, will help to significantly accelerate the time it takes OMass to identify lead compounds and bring life-changing benefits to patients.

“The professional and timely manner of this collaboration has enabled OMass to rapidly screen a novel target and help the company in its ambition to build a pipeline of small molecule therapeutics. The scale and speed of the platform is something OMass would not be able to replicate internally and we look forward to working with YellowDog in the future to further develop this capability.”

Giles Brown VP, Head of Medicinal Chemistry, OMass