What committees miss when they judge merit Diversity and equity policies are often dismissed as a numbers game: too many white men at the top, so institutions try to balance the…

I wrote a short piece on what funding and hiring committees often overlook when assessing merit.

@academic-chatter.bsky.social

✨New post ✨

Here I describe how I make maps in #ggplot2 from the GeoJSON, Shapefile, GeoPackage, and geodatabase formats 🗺️. I'll go a bit into projections and data wrangling of geospatial data. Hope it is of some use to some! 😊

danielroelfs.com/posts/advanc...

Link to the original thread that inspired this post: bsky.app/profile/osca...

The figures in the original report are clear, consistent, and professional so I wanted to discuss what their reasoning was and how to apply it elsewhere 😊

✨New blog post✨

I break down the beautiful data visualizations from a New America report (cowritten by @oscarp.bsky.social) and try to recreate them in R using ggplot2.

danielroelfs.com/posts/new-am...

Elsevier and Springer Nature had profit margins in 2023 that rival those of large tech companies.

Open access publication fees for some of the biggest publishers (Range is $200 to $12290)

My students and average people are often surprised to learn that academics pay a fee to publish a open-access paper in a journal, and _always_ surprised to learn that those fees can be as high as $12000 USD. #acadsky

danielroelfs.com/posts/the-mo... @danielroelfs.com

(bonus points for being open source and not being Microsoft Office 🙌🏻)

I absolutely adore reveal.js. I've used it as my main slideshow tool for the past years. If you know a bit of HTML/CSS it's super easy to create beautiful and dynamic slides (+ Quarto integration <3). I wrote a short post about my enthusiasm here: danielroelfs.com/posts/ode-to...

Plot showing a time series for 3 datapoints, two cases (where vineyard replanting happened at different timepoints in 2018 and 2021) and a control plot

Using publicly accessible data from the EU's Copernicus satellite project I did some (very) simple analysis on vineyard health over the course of a few years and showed how one might use it to measure recovery after a plot is razed due to illness or other planned or unplanned damage to the plants 🌱

✨New blog post✨

Remote sensing of vineyard health 🍇

I felt inspired by a data visualization exhibition I visited a while ago and decided to try my hand at remote sensing and analyzed some satellite images in Python.

danielroelfs.com/posts/remote...

What’s New in R: April 28, 2025 Explore the latest in R, including a new package for U.S. elevation data, updates in R 4.5.0 like the use() function, and a guide to creating FT data viz.

This week on What's New in R:

✅ A new R package by Joey Marshall for accessing elevation data
✅ Overview of what’s new in R 4.5, compiled by Russ Hyde of Jumping Rivers
✅ A guide to replicating Financial Times-style data visualizations, by Daniel Roelfs

Read the issue: buff.ly/Y4HwU4g

#rstats

I admire John Burn-Murdoch's ability to create really effective data visualizations. So much that I took the liberty of breaking down one of his recent figures that made the rounds and see if I could recreate it in R using only ggplot. Here's the breakdown:

danielroelfs.com/blog/financi...

Type of GitHub tile plot using Strava running data where each tile represents the kilometers run during the day. Sparse data with some data visualisation issues around the beginning and end of the year due to overlapping week numbers and because not all years start on Monday

It also let's us create some new nerdy plots like a GitHub tile plot, but for runs across the year:

Cumulative plot showing the cumulative kilometers ran during the 2022 (in yellow), 2023 (in pink), and 2024 (in 2024)

This let's you (re)create some analysis that is only available for subscription users and create plots that aren't available in Strava at all

Anyone that uses Strava has access to a large, self-generated (!), dataset 🧡. And at the end of the year it's quite fun to try to parse something interesting and relevant from that dataset: danielroelfs.com/blog/strava-...

Don’t believe the hype: AGI is far from inevitable | Radboud University Will AI soon surpass the human brain? If you ask employees at OpenAI, Google DeepMind and other large tech companies, it is inevitable. However, researchers at Radboud University and other institutes ...

See also www.ru.nl/en/research/...

GitHub - MichelNivard/awesome-complex-trait-genetics: A list of awesome tools for complex trait genetics. A list of awesome tools for complex trait genetics. - MichelNivard/awesome-complex-trait-genetics

🚨 This will become a curated list of awesome tools for complex trait genetics, **add yours**! it may become a review in which case those who contribute are invited as co-authors.

The slide decks were optimised for the screen in the auditorium (which was 1920x1200) so the slides are best viewed on a screen of that size, or perhaps on another 16x10 screen.

Slides for the trial lecture and disputation are available here: slides.danielroelfs.app/2024-11-19-p... Perhaps some of my illustrations in either slide deck are useful for others in the imaging-genetics or psychiatric genetics field! 😊

Photo of myself presenting at the trial lecture with a slide on twin-based studies in the background

Photo of myself during the disputation with a "thank you" slide in the background

Last Tuesday I defended my PhD in computational neuroscience and psychiatric genetics 🥳

PhD theses are rarely read outside of the supervisors and committee, so I tried to write it in an accessible way and published a digital version here: thesis.danielroelfs.app

- Daniel Roelfs, PhD

Mendelian randomization users, what should I call a polygenic score created for and used with MR?
I find PGS/PGI confusing to newcomers as not all ways of creating a PGS will do.
Would it be clearer to call this score an allelic score instead? Or something else?

Elsevier's parent company RELX has spent about as much money lobbying US politicians in 2023 as large financial companies like JPMorgan Chase and massive car companies like Ford www.opensecrets.org/federal-lobb...

Barplot showing the profit margins for a number of scientific publishers and other companies for comparison

Scientific publishing (Elsevier and Springer-Nature specifically) has higher profit margins than some of the biggest tech companies such as Apple and Google (up to 38% 🤯) . These margins have been stable for years

Open access fees for some of the biggest journals have exceeded $11.000 (120.000 NOK) PER ARTICLE.

It's hard to explain why scientific publishers like Elsevier and Springer-Nature make the kind of money they do.

So hard that I did the data analysis and wrote it down so I could easily share it and perhaps help others understand how broken the system is:

danielroelfs.com/blog/the-mon...

Preprint available here: www.medrxiv.org/content/10.1...
Data and code available here: github.com/norment/open...

The findings in this paper provide some additional genetic support for the dysconnectivity hypothesis in a number of psychiatric disorders as well as identifying potential targets for future etiological studies

This allowed us to find biologically relevant loci (involved in synaptic development and functioning) and discover shared genetic architecture with a number of psychiatric disorders. These overlapping loci are in turn also associated with a number of biologically relevant biological processes

fMRI has quite a lot of noise to say it kindly, so deploying a multivariate approach to look at global resting state functioning instead of individual networks really helped to capture relevant genetic variance.

Genetic overlap between multivariate measures of human functional brain connectivity and psychiatric... Using a multivariate approach in a sample from the UK Biobank, the authors examine the genetic relationships between measures of functional connectivity and psychiatric disorders.

Our latest paper is now published in Nature Mental Health ✨

We applied a multivariate GWAS implemented through the Bayesian framework in MOSTest on UK Biobank fMRI data and found widespread overlap with psychiatric disorders

www.nature.com/articles/s44...