Making data and analytics FAIR has transformative potential within organizations to build on existing knowledge. FAIR resources also democratize access to information and tools in underserved communities. Global standards and analysis platforms provide strong foundational elements. However, FAIRness across time and different sectors of the biomedical workforce presents challenges. Here we summarize how platforms make data and analysis FAIR today and what we see as key areas of future focus.
BioIT 2024 invited talk.
Building on a FAIRly Strong Foundation to Connect Academic Research to Translational Impact
1. Building on a FAIRly Strong
Foundation to Connect
Academic Research to
Translational Impact.
Jack DiGiovanna, PhD
Chief Science Officer
15 April 2024
4/23/24
Image credit: Susan Gregurick (2023)
2. How to talk to future humans?
2
Nuclear semiotics plans for 10,000 years into the future when languages do not remain
Image credit: https://hyperallergic.com/312318/a-nuclear-warning-designed-to-last-10000-years/
3. How to reuse our model in a grant proposal?
3
Moving from the bench to bedside
Reuse the data
& analytics?
Feb 2024
Publication April 2016.
Analysis in 2015
Search
usual external
hard drives
EPFL linked
Evernote and
NAS
Search
personal
Google Drive
Do old
laptops still
boot up?
Random walk
through other
hard drives
Wait, didn't I
use Dropbox
as a post-doc?
FOUND THE
DATA &
CODE!!!
5. FAIR analytics are crucial to drug discovery.
5
Basic research to standard of care spans decades and multiple stakeholders
Government &
Academia
Biotech Pharma
Regulatory
Agencies
Health
Systems
6. 10.5 years from Phase 1 to regulatory approval.
6
Significant time and changes in stakeholders before Phase 1
Image credit: https://www.ncardia.com/, thanks to Jeff Bissen
7. 10.5 years from Phase 1 to regulatory approval.
7
Significant time and changes in stakeholders before Phase 1
Image credit: https://www.ncardia.com/, thanks to Jeff Bissen
>5,256,000 minutes
Multiple teams have collaborated
Biopharma market has boomed &
busted – probably 2 reorgs
OS and analysis systems incremented
multiple major versions
How easy is it to package
the data & analytics for
review or sharing?
8. Models & standards help keep data FAIR.
8
Data governance lays the foundation for sustainable FAIR data
Common
Data
Elements
Data
model
definition
Data
Stewards, Data
Access
Committees,
DUA
Allowlists,
GA4GH
Passports
GA4GH
DRS
Share/search
data via FHIR,
LinkML,
SPARQL, etc
Sustainable
& accessible
data stores
10. FAIR analytics are more challenging.
10
Requires FAIR data, FAIR tools, and capturing experimental state
Interactive or low-code Production Pipelines
11. Platforms are a bridge to FAIR.
11
Analysis platforms include foundational elements that support FAIR analytics
12. Velsera platforms support organizations holistically.
An enterprise-grade platform … … providing the flexibility and
compatibility you need …
… and enabling the right
collaboration for the right outcomes
Public app gallery Collaborative Research & Development
Access 900+ ready-to-use optimized tools and
workflows that also can be incorporated into custom
pipelines
Collaborate within and across organizations while
keeping control of your assets with fine-grain
permission level
Connected cloud De-siloed data management
Keep data in your own buckets with storage support
for AWS, Azure, and Google Cloud Platform
Out-of-the-box capabilities and extensive metadata
infrastructure to handle large data and unlock
disparate data sets
Industry-standard tools for reproducibility and
compliance
Experienced Professional Services team
Multi-cloud
All workflows can be run on multiple cloud vendors
and regions of your choice, enabling you to analyze
data where it lives
Ease of pipeline migration
Easily bring your existing tools and workflows
in and out of the platform
Workflow editor for customization
Leverage the drag-and-drop interface to build
and customize new pipelines
Build, customize, automate, and execute analysis
pipelines using established tools, APIs & CLI in secure,
compliant & regulated ISO and FEDRAMP environments
Experienced service team that provides clients the
end-to-end support they need to manage their
genomic and phenotypic data ecosystem
12
13. Interoperability & standards are baked in.
13
Faster research: Streamlined data sharing accelerates
discovery, leading to quicker treatments and improved
patient outcomes.
Empowered users: Seamless access to relevant data allows
for personalized medicine and more informed decisions.
Democratized data: Open standards break down data silos,
fostering inclusivity and collaboration across the ecosystem.
Velsera's platforms & commitment to open standards drives FAIR innovation
DRS
GA4GH
Passport
FHIR
Files
AuthN/Z
Clinical
Metadata
Data
14. Interoperability & standards are baked in.
14
Faster research: Streamlined data sharing accelerates
discovery, leading to quicker treatments and improved
patient outcomes.
Empowered users: Seamless access to relevant data allows
for personalized medicine and more informed decisions.
Democratized data: Open standards break down data silos,
fostering inclusivity and collaboration across the ecosystem.
Velsera's platforms & commitment to open standards drives FAIR innovation
DRS
GA4GH
Passport
FHIR
Files
AuthN/Z
Clinical
Metadata
Data
15. Standards also help "package" analytics.
15
Directly export production pipelines and related experimental context for regulatory review
"BioCompute
Object"
IEEE 2791-2020
Regulatory
agency
CWL is explicitly reproducible (order of operations specified
by graph), vendor neutral, & has transparent governance.
16. Case study: Pharma immunotherapy clinical trial.
16
CHALLENGE
ü Secure, performant workspace to store, analyze, and jointly interpret data in a clinical trial setting
ü Versioned, GxP, 21 CFR Part 11- compliant workflow to support trial
ü Automated processing for error free data handoffs and maximum efficiency
ü Scientific expertise to ensure the most optimal outcomes
RESULTS
• Benchmarking of tools and custom bioinformatics workflow development
• A centralized data hub to streamline data management
• Automated code for orchestrating data management and processing
• Compliant SOPs followed by Velsera SMEs
SOLUTION
Need to efficiently manage data logistics with multiple vendors
and handle the increasing complexity of bioinformatics
analyses to support an immunotherapy clinical trial.
17. FAIR analytics also sparks collaboration.
17
Platforms help share insights and create competitive advantages for your teams
18. Key remaining challenges to FAIR analytics.
18
Image credit: https://xkcd.com/927
Time takes
everybody out,
it's undefeated.
- Rocky Balboa
19. Key remaining challenges to FAIR analytics.
19
Image credit: https://xkcd.com/927
Time takes
everybody out,
it's undefeated.
- Rocky Balboa
20. Yes, we can overcome those challenges.
Invest (early) in solid Data Governance
Support sustainable data stores, persist through changes
Provide consistency in AuthN/Z
Support crucial tool and model repositories
Engage with the standards community; advocate for
evolution of the specs to support FAIR analytics
Engage with workflow description governing bodies to
ensure reproducibility
Adapt with technology, analytics, & regulatory agencies
4/23/24 20
21. FAIR investments accelerate drug discovery
& support collaboration.
21
We should continue evolving our focus to invest in Data & Analytics as a Product.
An interoperability triangle of object pointers (GA4GH DRS), AuthN/Z (GA4GH passports), and metadata (e.g. FHIR) is
a great foundation for FAIR data.
Containerization and workflow languages (partially) support FAIR workflows.
Additional standards help create FAIR analytics; integration with underlying Platforms captures key information
(e.g. configuration, references, provenance, system state, tool information).
Continued support from our community is crucial to keep data & analytic resources achieving their full potential.
Booth #704