Presentation from PEARC20 (Practice & Experience in Advanced Research Computing) by Hertweck and Strasser, published article here: https://dl.acm.org/doi/abs/10.1145/3311790.3396655
Training and documentation for on-premises infrastructure represent the foundation of most institutional support for computational researchers. For most academic research institutions, however, these approaches fall short of meeting the needs of diverse researchers with different levels of experience with data-intensive research. We describe a framework for characterizing levels of computational expertise and relate this model to informational support provided for biomedical researchers at a non-profit/academic research center. Our model differentiates between novice, competent practitioner, and expert users of reproducible computational methods, and is related to the composition and needs of an entire research community. We specify methods best suited for researchers with different levels of expertise, including formally structured short courses, code examples/templates, and online wiki-style documentation. We provide recommendations to encourage the development and deployment of these resources, and suggest methods for assessing their effectiveness. Supporting multiple types of informational resources for researchers with different computational needs can be labor-intensive, but ideally increases computational ability for the entire institution.
From Novice to Expert: Supporting All Levels of Computational Expertise in Reproducible Research Methods
1. From novice to expert:
Supporting all levels of computational
expertise in reproducible research methods
Kate Hertweck and Carly Strasser
Fred Hutchinson Cancer Research Center
@k8hert
2. How do I do [some task]?
Kate Hertweck (@k8hert), Supporting all levels of expertise
realityexpectation
help(print)
4. Kate Hertweck (@k8hert), Supporting all levels of expertise
thecoop.fredhutch.org
The Coop Community
Biomedical and clinical
research experts
Variable acceptance of
open science principles
Customizable and flexible
solutions required
5. Kate Hertweck (@k8hert), Supporting all levels of expertise
Novice
Competent
practitioner
Expert
Application of
computational
skills to own
research
Routine application
of skills and
adoption of best
practices
Levels of computational expertise
7. Kate Hertweck (@k8hert), Supporting all levels of expertise
What are the pressure points for adoption
of reproducible computational methods?
What do people need to learn?
+
How are they most likely to learn?
8. Kate Hertweck (@k8hert), Supporting all levels of expertise
Types of support: short courses
fredhutch.io/resources
carpentries.org
Materials adapted to
suit the specific needs
of our community
9. Kate Hertweck (@k8hert), Supporting all levels of expertise
Types of support: code examples/templates
Application of code to
common workflows
Code that works, but is
also aligned with best
practices
10. Kate Hertweck (@k8hert), Supporting all levels of expertise
Types of support: demos and tutorials
Bridge between reference
documentation and
developing code for a
specific project
11. Kate Hertweck (@k8hert), Supporting all levels of expertise
Types of support: SciWiki
sciwiki.fredhutch.org
13. The usefulness of training and
documentation depends on both what
information is provided, and how it’s
delivered.
Kate Hertweck (@k8hert), Supporting all levels of expertise
Learning technical skills is “easy.”
Learning to apply technical skills to your own
research problems is much more difficult.