Shadolith

the brave little choster

audio things for Deltarune + Homestuck + Hiveswap + other stuff | Xenoblade my beloved


celechii
@celechii

holy shit i am learning that there is a worm that is 1mm long called the caenorhabditis elegans that has so few cells in it's body that it's known and documented what EACH of the <1000 CELLS is and does??? it's brain only has 302 neurons which means apparently they're not far off from being able to record this thing's complete neural activity.

where this is going is a project called OpenWorm which looks like an attempt to digitally simulate the entire physical+neural workings of this worm??

here's the OpenWorm Wikipedia and their website


folly
@folly

tumblr worm fandom: mean girls talking about amy dallon
cohost worm fandom: hey what if we sequenced every neuron in a worm's brain. this is normal to want and possible to achieve


wick
@wick

Specific loops of neurons that output a rhythmic behavior are called central pattern generators and they're the shit. We have them too for stuff like breathing and walking and chewing! Vertebrates have big groups of neurons acting together instead of individual cells, but they're organized similarly.

(The game is Crescent Loom, it works in-browser.)


erica
@erica

how are people this cool


You must log in to comment.

in reply to @celechii's post:

I did volunteer work on a project last year that was working on doing the same thing with a fruit fly's brain, which is about 400 times bigger. We actually published results earlier this month! https://www.biorxiv.org/content/10.1101/2023.06.27.546656v2.full

I've been volunteering for these projects for a decade this May and it's been like my personal introduction to the growth of AI. Originally it was a giant 3D coloring book with areas the AI was pretty sure was one complete object that were maybe 10 nanometers (!) a "side" (they were maybe cylindrical if you were lucky but often just messy blobs) and while it was still a daunting task for a collaboration of labs to accomplish it became a simple enough task that you could distribute to volunteers and they could map out entire neurons within a day or two. For reference the experts charting c. elegans took more than a decade for its 302 neurons. They were also like coloring in 2D pictures by hand, let alone stacking several on top of each other to get a 3D shape. Now, though, the AI is able to get near-complete pathways of entire neurons (but will still make fuckups that have to be checked by a person), but you're talking about getting neurons finished by 1 person in 20 minutes, or a couple hours if it's messy.

It's also largely outpaced the growth of the citizen science projects that accompanied this (initially looking at tiny slices of mouse brains) that I think all of the open projects people can work on themselves exist only on leftover data people have largely moved on from? Which is a little sad but it's not like this is a completely closed door to volunteers and it's also like the most ethical use case for AI that I can think of because it's replacing 0 jobs just taking tedious work away from volunteers who themselves were taking it away from tired grad students.

in reply to @ldx's post:

in reply to @wick's post:

in reply to @erica's post:

Pinned Tags