What was the impact of the AI solution?

The team completed indexing in six days instead of what would have previously taken years. Cost per dataset fell from two or three dollars to about five cents. An initial pilot uncovered over a billion versions of plastic-degrading enzymes in ten hours.

Source

Logan: Unlocking nature's secrets to fight microplastic pollution

Use case typeData platform modernizationUpdated 5 days ago

Artem Babaian and the University of Toronto’s Donnelly Centre Laboratory for RNA-Based Lifeforms built Logan, an open-source searchable index of the world’s public DNA/RNA sequencing data, to make massive genomic datasets searchable at global scale. The team used AWS infrastructure to process and index 39 million sequencing datasets, then applied Amazon Bedrock models to identify which environments protein sequences are found in and shortlist promising plastic-degrading enzyme candidates. The workflow aims to accelerate discovery and testing of enzymes that could help address microplastic pollution and other scientific problems.

Organization: University of Toronto
Industry: Healthcare
Location: Canada
Published: July 2026

Reported outcomes

1,000,000,000 versions

enzyme versions discoveredTime & speed

6 daysindexing time5 USD cents per datasetdataset processing cost

Strategic outcomes

Scale & capacityMade the world’s public sequencing data searchable at global scaleSustainability & ESGAccelerated plastic-degrading enzyme discovery for microplastic pollution

Primary read

Use case focus

Showing 2 of 2

1Data platform modernization
2Drug discovery

Massive public DNA and RNA sequencing data were too large to search efficiently, creating a bottleneck for finding useful enzymes for microplastic degradation.

The team deployed Logan on AWS and ran 2.2 million CPUs in parallel over several days to index sequencing data in six days.
They optimized their cloud workflow on AWS to reduce dataset processing cost from dollars to cents per dataset.
They then used Amazon Bedrock models to analyze protein sequence environments and prioritize candidates for laboratory testing.

Technologies

Amazon Bedrock AWS Open Data Registry AWS compute

The team completed indexing in six days instead of what would have previously taken years.
Cost per dataset fell from two or three dollars to about five cents.
An initial pilot uncovered over a billion versions of plastic-degrading enzymes in ten hours.

Sources & evidence1

Groundedness: 5/5Type: Customer StoryPublished: Jul 8, 2026Publisher: AWSEvidence: PrimaryConfidence: High

Primary source

AI-generated summary. Verify important details with the linked sources before relying on this case.

Explore related AI use cases

Customer evidence