Michael Clerx
@michaelclerx.com
660 followers 930 following 1K posts
I do sums to learn why heart cells beat. Computational biologist. Biological computer. Sometimes does experiments. Whippet owner. Makes bad jokes. He/him.
Posts Media Videos Starter Packs
Reposted by Michael Clerx
nikitagill.bsky.social
The most annoying thing about the AI bubble is we can all see it happening, we all know it’s going to burst and we all know what the consequences will be. Everyone except apparently, Big Tech and the government.
michaelclerx.com
I, um...
I went on a website - not a dodgy one - right? And
Well
And it showed me an ad
An advertisement showing six panels: two containing what looks like human intestines, one with a five-sided star shaped intestine (?), a lady with a bridal veil, a lady in costume (Hawaiian?), and the mum from Bluey
Reposted by Michael Clerx
thetnholler.bsky.social
ICYMI: “And now, a response from the leader of the frog resistance...” 🐸 🎵 #Colbert
michaelclerx.com
Mark "Spooky" Watson
Reposted by Michael Clerx
govindajeggy.bsky.social
Arachnophobia is the fear of anything with eight legs, like spiders, scorpions or ABBA
Reposted by Michael Clerx
scratchcarddust.bsky.social
If you know basic kabbalah it's actually not too hard to make a golem out of a labubu. Don't recommend it though. They have a pre-existing body-dharma. As soon as they wake they run straight for the sea. V hard to counter. Lost the first one. No idea what it's doing out there. Moved like lightning.
Reposted by Michael Clerx
natcassidy.bsky.social
This photo of Robert Bloch is the most writer photo to have ever writer photo’d. It is Pure Writer Photo.
michaelclerx.com
Dare to ask
ceej.online
any culture, at any point in history, could have invented darude’s sandstorm. the fact that they didn’t is just further proof that old times were busted generally
michaelclerx.com
Went to t sheep dog tribulations today.
Don't know about you but I just prefer m to the trials, you know? Feel a bit more real to me.
Reposted by Michael Clerx
michaelclerx.com
Additions to the list are HIGHLY welcome! Especially if you know of source code I missed
michaelclerx.com
So code sharing is definitely on the up in CCE, but we're not out of the woods yet!
michaelclerx.com
A more optimistic way of plotting the same data, shows that we've gone from ~95% no-code in 2000, to 30.2% of models published with procedural code in 2025, and 14.5% published with (even better!) declarative code.
michaelclerx.com
To make it pretty, I had to group a lot of formats under "Other". To get a better view of what "other" formats are, I made this graph excluding the big three (no code, matlab, cellml). From top to bottom it goes Myokit (yay!), C++, simBio, C, CARP, Delphi, Fortran, Visual basic, and Neuron
Myokit can be seen growing since 2016, C++ was popular around 2010, simBIO between 2005 and 2010. C is added slow and steady. The remainder doesn't show any interesting patterns.
michaelclerx.com
So with the parsing (and a standardised format in the markdown) in place, you can start doing some fun graphs!

Fun - but heavily caveated. This is based on what is on my list, not what exists out there. But I had a go at least 🤷. The entries outside of "No code" are hopefully a good lower bound
A graph showing "author-provided code" available online, for cardiac cell ephys "models" published between 2020 and 2025. The graph shows that in 2000, no code was online (or if it was those websites are long-since deleted), while in 2025 we have author-provided code for just under 45% of all CCE models ever. The graph is further split by code type, showing 17.9% of models have author-provided matlab, 9.5% cellml, and then it gets into 5% or less. Myokit is the 4th most popular, most likely reflecting my bias in collecting this data.
michaelclerx.com
I just recently got asked to talk at the CellML workshop in NZ (so late at night via zoom, no jet setting just yet), and then started thinking if I could maybe parse my list and do some graphs

(This is a lie. I started thinking THAT I could parse and do some graphs, and that this was a good excuse)
michaelclerx.com
So a while ago I got annoyed that I couldn't remember a particular cardiac cell ephys model, and thought there should be a list, so started making that list... github.com/myokit/model...

(Yes it's a bit silly, yes there's PMR and others, yes it's hard to say when it's a "change" or a model etc)
GitHub - myokit/model-list: A list of (detailed, non-stochastic) action potential models, with links to papers, source code, CellML and Myokit implementations
A list of (detailed, non-stochastic) action potential models, with links to papers, source code, CellML and Myokit implementations - myokit/model-list
github.com
Reposted by Michael Clerx
anandmenon.bsky.social
It's beyond me why the PM doesn't want to put out this simple message. It's not hard. Hold a reception for prominent non-white British sportspeople, actors, etc to point out the kinds of well known and well liked people this kind of ugly nationalism is attacking. It's really not rocket science. 5/5
Reposted by Michael Clerx
henrymance.ft.com
The right-wing campaign group that helped topple Roe v Wade in the US is now working to roll back abortion laws in the UK, with the help of Nigel Farage.

Its first step? Trying to create a debate around "free speech"
www.nytimes.com/2025/10/13/w...
michaelclerx.com
Is anyone working on LLMs that are not a service? I think that's one of the main things that puts me off. If I'm going to have a little robot pal that helps me with things, I want it to be like a puppy that lives in my house, not a terminal that feeds me "advice" from multinational corporations