Terrell Russell, Ph.D

Executive Director

iRODS Consortium

May 28-31, 2024

iRODS User Group Meeting 2024

Amsterdam, Netherlands

The Rapidly Emerging Antiviral Drug Development Initiative AViDD Center (READDI-AC) is an NIH-funded public-private partnership focused on developing effective antiviral drugs to combat emerging viruses.

The READDI-AC at UNC-Chapel Hill is one of nine Antiviral Drug Discovery (AViDD) Centers funded by the US National Institute of Allergy and Infectious Disease (NIAID) at the National Institutes of Health.

$65M in 2022 - 40 Investigators, 23 Research Sites, 5 Countries

NIH Award 1U19AI171292-01

READDI-AC

The response to viral outbreaks has historically been reactive – vaccines and medications are developed only after a new virus emerges. Our mission is to proactively prepare for emerging viruses by developing antiviral drugs that are active against more than one virus in a family. These broad-spectrum antivirals will help safeguard the well-being of communities worldwide against existing viruses and will be more likely to be effective against future novel viruses in the same family.

Four families:

• Coronaviruses - causes SARS, MERS, COVID-19
• Filoviruses - includes Ebola, Marburg
• Flaviviruses - includes West Nile, Dengue, Zika
• Alphaviruses - includes Chikungunya, Equine Encephalitis

READDI-AC Mission

RENCI, as a subawardee, was tasked to assess, design, and develop the data management solution for the READDI-AC project.

Timeline

• Interviews - July-August 2022
• Survey - August 2022
  • Determination of existing lab workflows
  • Document types, variety, size, volume
  • Number and identity of humans in the loop
  • Opportunities for automation
  • Opportunities for cross-lab interactions
• Security considerations - Fall 2022
• Initial design of system - Fall 2022
• Paper evaluation - Fall 2022
• Initial implementation - Nov-Dec 2022
• Testing - Dec 2022
• Deployment - Jan 2023
• Evaluation - Q1 2023
• Iteration - through 2023

• If you use instruments in your work, what is the format of files they produce?

• Where do you currently record and store chemical or biological data? In what data format?

• What keywords or other terms do you use to search for previously recorded data?

• Do you use existing or public vocabularies, ontologies or other references?

• Are you familiar with FAIR data sharing principles?

• What are typical dataset sizes for each unit of work?

• What is a typical data generation rate for your lab / unit?

• What is a typical number of data files you generate per week?

• Do you protect the stored data (do you require secure login/authentication to access your data?

• How do you currently share your data with (i) your labmates, (ii) within UNC, (iii) with the rest of the world?

• What software do you use to process or visualize your data?

• Do you have a need to format your data for publication or presentation format?

• What steps are manually processed today? What steps are automated today?

• Where is manual processing required? Where can data processing be automated?

• What limitations are your lab / group / team running into?

• What is your highest priority or need for data capture or storage?

Discovery Questions

• Not Big Data (yet)

  • 1000s of files over the course of a year

  • Maybe 10s of GBs, but most much smaller

  • Human scale rate of ingest

• A few formats, mostly open / convertible

  • doc, pdf, xls, ppt, csv, txt, prism, jpg

• Some electronic lab notebooks

  • Mostly xls

  • Manually calculated / generated

• Very little currently automated

  • Manual transcription from paper notebooks

  • Graphing done in Excel or (rarely) Jupyter notebooks

Discovery Findings

• No shared naming conventions

  • ​For either data filenames or metadata

  • Sometimes consistent within a lab

    • ​Due to necessity

    • But not over time

• ​​Highest priority is access / sharing

  • ​​This project's data needs a 'home'

• ​No centralized data repository

• No standardized data processing (raw to publication quality) and data upload protocols

• No versioning protocols

• Metadata currently non-existent

• Negative results / Failed attempts not recorded

• HEIGHTENED NEED for RDM due to new NIH reqs

There was very little process to automate - we were starting from scratch and these labs did not have much in common.  Different instruments, different chemistry, different software, different processes, different formats.

Not their fault - they'd never been required to coordinate and collaborate in the past except via publications.  This was a new mandate.

There would be two projects:

• People engineering
  • hardest part, scientists do not want to change their processes
  • requires many people to coordinate (expensive in time and effort)
• Software engineering
  • a few puzzles to solve, but nothing too daunting
  • security requirements demand working with other parties

Design and Preparation

Software Engineering Requirements

• federated login for otherwise unaffiliated researchers
• secure enclave
• just files, mostly spreadsheets
• some annotation
• automation where possible
• search
• available for analysis with existing tooling
  • probably via download

January 2023 - Initial Deployment

Angular Application

• upload
• assays
• search
• compound profile
• FAQ
• profile information

iRODS Policy - Four recurring rules

• irule davidd_add_sweeper_to_queue
• irule davidd_add_compound_profile_sweeper_to_queue
• irule davidd_add_compound_profile_remover_to_queue
• irule davidd_add_assays_sweeper_to_queue

DAViDD - Application and Policy

irule davidd_add_sweeper_to_queue

• davidd_find_and_parse_uploaded_files
• davidd_parse_and_place_jsonfile
  • parse python dict
  • prepare avus_to_add
  • decode file data, write it
  • associate avus

February 2023 - Upload

March 2023 - File Metadata

GenQuery

• matching on file name and metadata

March 2023 - Search

irule davidd_add_compound_profile_sweeper_to_queue

• davidd_process_requested_profile
• davidd_process_queued_file
• davidd_walk_collection_for_compound_info
  • use openpyxl, read spreadsheets, populate new one

irule davidd_add_compound_profile_remover_to_queue

• davidd_remove_old_compound_profiles
  • defined by compound_profile_removal_age_in_minutes

September 2023 - Compound Profile

irule davidd_add_assays_sweeper_to_queue

• davidd_find_and_parse_assay_files
  • davidd_parse_and_place_jsonfile

January 2024 - Assays

Discovery and prototyping were a success

• 4 labs interviewed
  • Many challenges identified and lessons learned
• 3 federated login architectures attempted
  • Selected CILogon.org

353 datafiles uploaded in the first year

• 105 Coronavirus
• 173 Alphavirus
• 27 Filovirus
• 48 Flavivirus

Summary

Having identified the main requirements and bench-to-data process, the project selected an existing commercial vendor for its extensive GUI and compound-specific analysis tooling.

RENCI continues to develop database-level tooling focused on chemical compound information and linkages with other tools in the ecosystem.

Acknowledgements

• NIH
• Ava Vargason, Nat Moorman, Ralph Baric, Tim Willson, Toni Baric
• Oleg Kapeljushnik, Kory Draughn, Alex Tropsha, Robert Hubal, Kelyne Kenmogne, Carrie Pasfield, Patrick Patton

The Future

Thank you!

Questions?