The challenge
A global crop science organization’s trait discovery program had outgrown the infrastructure supporting it. Years of organic growth had left the R&D organization running a patchwork of internal platforms, including an internal discovery portal, a BLAST Portal, a Knowledge Base, and specialized tools for marker analysis, phylogenetics, and toxicity screening. Each had been built to solve an immediate need. None had been built with the others in mind.
The result was familiar to any genomics organization operating at scale: fragmented data stores spanning MongoDB, MariaDB, Oracle, MySQL, PostgreSQL, and Elasticsearch, a genomics pipeline environment tied to an aging on-premises cluster, and no consistent way to discover or reuse existing tools across breeding, gene editing, and pangenome analysis teams. Scientists doing molecular breeding, genetic mapping, and cheminformatics work were often unaware that tools already existed to answer their questions, or were duplicating pipelines other teams had already built.
Compounding the problem, the organization’s genomics infrastructure spanned an on-premises cluster with petabyte-scale storage and a growing set of AWS workloads, with no unified strategy connecting the two.
The approach
BioTeam ran a structured enterprise architecture assessment across the trait discovery platform, starting with an AS-IS review of the existing environment. This meant cataloging every tool in active use, from BLAST search and genome comparison utilities to internal plant phylogenetics and phytotoxicity screening tools, and mapping how sequence, variant, and marker data actually moved between them.
With the AS-IS picture established, BioTeam developed a TO-BE architecture and roadmap addressing three layers of the problem at once:
Workflow modernization. BioTeam recommended standardizing pipeline development around Nextflow and Galaxy, replacing ad hoc scripts and custom SSH job submission with reproducible, portable workflows suited to genome assembly, ortholog search, and trait-marker association work.
Infrastructure strategy. Rather than a wholesale lift-and-shift, BioTeam recommended a hybrid path: keep compute-intensive breeding and sequencing workloads on the existing on-premises cluster while moving metadata management and selected Nextflow workflows to AWS, where they could take advantage of S3 and Glacier for lifecycle-based storage of large-scale sequence and variant datasets.
Organizational readiness. Infrastructure change without staffing and governance change rarely sticks. BioTeam’s roadmap included hiring recommendations for two DevOps engineers, a data curator, and a business analyst, alongside the formation of a Data Management Working Group to own metadata standards and documentation practices going forward.
The solution
The resulting architecture gave the trait discovery organization a clear path from a fragmented tool landscape to a unified sequence datastore, with:
- A consolidated approach to portal and tool discoverability, so scientists could find and reuse existing capabilities across internal discovery portals, BLAST, and knowledge base tools instead of rebuilding them
- Recommended Nextflow templates and workflow best practices for standardizing genome assembly, variant analysis, and gene annotation pipelines
- A hybrid cloud model pairing the on-premises cluster with AWS for metadata and select workflow execution
- DevSecOps improvements, including SSL implementation, to bring the environment up to modern security standards
- A staffing and governance plan to sustain the new architecture beyond the engagement itself
The outcome
The organization now has a validated roadmap for migrating trait discovery workflows to AWS, with metadata infrastructure already underway in transition. The assessment gave R&D leadership a clear, sequenced plan for closing the gap between where their genomics infrastructure stood and where a modern, multi-region trait discovery program needs to be, spanning tool consolidation, cloud strategy, and the people needed to run it.
