iiinova — where your digital self keeps evolving

James Thomas

@jatonline.uk

179 followers 509 following 42 posts

Head of Research Engineering at the Jean Golding Institute for Data and AI, University of Bristol. jatonline.uk

Posts Replies Media

James Thomas @jatonline.uk · 12/08/2025

@buttondown.com Assume you get notified about 500 errors, but should I report any other way?

Screenshot of Buttondown website's "Confirm Unsubscribe" page.

Screenshot of Buttondown website showing a "500 Internal Server Exception" page.

120

James Thomas @jatonline.uk · 04/05/2025

Last week I gave a lunch and learn talk on "Can others reproduce your conda results? Using 'pixi' in a Python data analysis project" showing how keeping track of package version using conda is hard and can go wrong if your developer use multiple platforms, and how pixi helps.

Title card for the video: Can others reproduce your conda results? Using pixi in a Python data analysis project. By James, Thomas, Jean Golding Institute, University of Bristol

121

James Thomas @jatonline.uk · 30/04/2025

This may be minor and/or unrelated and/or not a problem, but it looks like you have a ~ character in your path that isn't being expanded to your home directory in the output lower down. If you're using ~ in the path that points to your python interpreter, you may need to use $HOME instead.

VS Code error messages showing:

Process Execution:
~/anaconda3/envs/defaultwork/bin/python

09:12:06.174 [error] Failed to request kernel info [Canceled: Canceled at Ah. value /Users/~/. vscode/extensions/ms-toolsai.jupyter-2025 3

100

James Thomas @jatonline.uk · 22/04/2025

👀 TIL about the Wagtail and Django release timeline: github.com/orgs/wagtail... Essential reading, alongside the Django and Wagtail release schedules: www.djangoproject.com/download/#su... github.com/wagtail/wagt... Thanks for this @wagtail.org!

Screenshot of the Wagtail and Django release timeline, filtered for releases prior to June 2025, with security support at least until today (22 April 2025). Shows Wagtail 7.0 LTS (in the future), and current supported versions (in order): Django 5.2 LTS, Wagtail 6.4, Wagtail 6.3 LTS, Django 5.1, Wagtail 5.2 LTS and Django 4.2 LTS. Wagtail 5.2 LTS support ends soon.

000

James Thomas @jatonline.uk · 08/04/2025

Tried out some live coding 😬 but it went ok! The Lunch and Learns are part of our training in research software engineering and data science. More lunch and learns here: bristol-training.github.io/lunch-and-le... and our full training material is here: bristol-training.github.io/courses.html

Screenshot of a terminal session. The terminal says:

Project was started with: uv init -- lib

$ tree

* README. md
* pyproject. toml
* uv. lock
* notebooks
* example_notebook.ipynb
* src
* my_project
* __ init __. py
* probability.py

4 directories, 6 files

130

James Thomas @jatonline.uk · 12/03/2025

To finish off the session, Will Chapman, Data Scientist at @jgibristol.bsky.social seeks wisdom from the room on the hidden challenges in reproducible health data research... how can we have an institutional memory of these, and not have to rediscover them with each new project?

Will Chapman presenting a slide at the Health Data Research Network. The slide says:

Purpose of this session

We want to understand the current processes) for getting routine data for research from our local NHS
- "Easy data" that is fast to specify and extract
- Structured data that needs transformation before transfer
- Multi-modal data that can't be created through the right query, eg images, vitals

We want to understand what it took to get (routine) data for health research via the Trusts
- Which parts of the process went well, with good support from UoB/NHS?
- Which parts of the process were lengthy or painful?

This will feed into a project to streamline health data pipelines for both the SDE and our usual research platforms

021

James Thomas @jatonline.uk · 12/03/2025

Ben Elsworth, Lead Data Engineer at Our Future Health ourfuturehealth.org.uk talked about their recent developments. "(probably) the most efficient and advanced genetic data processing and release pipelines in the world" & "VCF vs Zarr" for data storage "Zarr chunks the data [and is] parallelisable"

Ben Elsworth talks about research at Our Future Health at the Health Data Research Network. The slide says:

What our participants donate to Our future Health

Complete a baseline questionnaire (demographics & household, work & education, lifestyle, family history, personal health history)

Physical measurements taken at clinic appointment (height, weight, waist circumference, blood pressure, heart rate)

A blood sample for genotyping and for long-term storage of plasma, buffy coat, and DNA

DNA is genotyped using a custom array that captures 700k up-to-date variants of interest for optimal research insights

Provide permission for data linkage to health-related datasets (e.g., primary care, secondary care, cancer, death)

Provide permission to be recontacted for further surveys/biospecimens/assessments and to be invited to additional studies/trials

100

James Thomas @jatonline.uk · 12/03/2025

Richard Welpton, Head of Data Service Infrastructure at Economic and Social Research Council @ukri.org spoke about the key lessons for improving researcher experiences when working inside Trusted Research Environments (TREs). "Researchers should work collaboratively - TREs should support this"

Richard Welpton presenting a slide at the Health Data Research Network. The slide says:

Other key lessons

Researchers work collaboratively.
TREs should support this.

Invest in tech, invest in people:

Automate the repetitive and mundane.

TREs and researchers need to understand each other.
Invest in specialist support

100

James Thomas @jatonline.uk · 12/03/2025

Fiona Booth, Data integrity & Quality Assurance Programme Director at @bristoluni.bsky.social outlined the differences in what is understood by 'reproducibility' between commercial and academic health research. The most popular word in the room: documentation.

Fiona Booth presenting a live survey at the Health Data Research Network. The slide says:

What do you think makes research reproducible?

published methods
methodology is robust
open code
access
honesty
accessible code good code base
publications
readability
visibility
clear datasets
open data
open reporting access to data
documentation
curated data
robust data sets
version control
open source
consistently
must be desired
openness
code
access to code
transparency
reusable script
accurate reporting
sharing code
common methods detailed meta data
method is available

100

James Thomas @jatonline.uk · 12/03/2025

Damon Chow, Business Data Analyst at HDR UK gave a live demo of the Health Data Research Gateway website healthdatagateway.org, demonstrating the search functionality and how you can find information on datasets, data custodians, and more.

Damon Chow giving a live demo of the Health Data Research Gateway website at the Health Data Research Network.

100

James Thomas @jatonline.uk · 12/03/2025

Rachel Denholm, Senior Lecturer, @bristoluni.bsky.social Medical School, and Co-Director of Health Data Research UK South West, kicked off the event with an outline of where we are now with secure data environments (SDEs or TREs) and what people are working towards, including accessing compute.

Rachel Denholm presenting a slide at the Health Data Research Network. The slide says:

What is a secure data environment (SDE?)
(= trusted research environment (TRE))

Data access process

Accredited researcher proposes project
→ Request approval from one advisory group
→ Access via a remote portal. Cannot copy or download data. Outputs exported only after checked by an independent researcher
→ End of project, access removed

100

James Thomas @jatonline.uk · 12/03/2025

More interesting talks coming up...

Programme for the Health data Research Netowrk workshop on 12th March 2025.

10.10 Rachel Denholm, Senior Lecturer, Bristol Medical School, UoB, and Co-Director of Health Data Research UK South West
The Rise of New English Secure Data Environments

10.30 Damon Chow, Business Data Analyst, HDR UK
The Health Data Research Gateway

11.10 Fiona Booth, Data integrity & Quality Assurance Programme Director, UoB
Commercial vs. academic research approaches to reproducibility

11.30 Richard Welpton, Head of Data Service Infrastructure, Economic and Social Research Council
Improving the experience of accessing sensitive data for research

11.50 Ben Elsworth, Lead Data Engineer, Our Future Health
Processing and releasing genetic data at Our Future Health

12.20 Will Chapman, Data Scientist, Jean Golding Institute
Routine data for health research

12.40 Q&A and round table
Challenges in health data research: aims for the HDRN

100

James Thomas @jatonline.uk · 12/03/2025

My colleague Pau Erola has organised a great workshop to share knowledge on securely using health data in research. Part of the Health Data Research Network #HDRN: hdrnet.github.io supported by @jgibristol.bsky.social & Research England Enhancing Research Culture funding via @bristoluni.bsky.social

Pau Erola presenting a slide at the Health Data Research Network. The slide says:

What will we do?

The HDRN will organize workshops every month that will discuss about specific topics on health data science. The events will be an opportunity to share knowledge, discuss and agree best practices, and networking.

Taking inspiration from The Turing Way and Data Hazards, each workshop will be reflected in a chapter of a live, collaborative book on standard practices for health data science.

Damon Chow presenting a slide at the Health Data Research Network. The slide says:

Enhancing the Gateway, followed by a diagram showing the different parts of the Health Data Research Innovation Gateway from HDR UK. Including: 900+ metadata descriptions, Data Access Requests, Data Use Register, Advanced search, 4,000+ tools and publications.

A room of people watching a speaker talk in front of a screen, at the Health Data Research Network.

130