Stefan Dvoretskii

Neuroengineer interested in Biological Artificial Intelligence and Complex Systems.

[Effie Parks @OnceUponAGene]

Do you love the movie - Hackers?
Of course you do.
Maybe you'll enjoy this story too.

"She Found the Backdoor - One Mom’s Hack on Genetic Diagnosis w/ Julia Taravella"

I met Julia Taravella in a Seattle coffee shop in 2022, after she messaged me while she was in town from New Orleans. She’d been listening to my podcast. She told me her story so nonchalantly and glossed over the part where she taught herself genetics, wrote her own algorithms, and diagnosed her two boys because no one else would. She reminds me of one of the best movies of all time - Hackers. If you haven’t seen it, you really need to throw on some roller blades and get your priorities straight. Julia’s the rare mom version of this movie. A White Hat with zebra stripes - obviously. She's calm, deliberate and like all determined rare parents, a bit dangerous.

Julia didn’t set out to be a genetic sleuth. But when doctors dismissed her sons’ developmental delays as 'just autism,' she knew she had to figure it out herself. Instead of presents she ordered consumer genetic tests as Christmas presents for the whole family back when that kind of thing was still fringe and very expensive. She downloaded the raw data most people didn’t even know was there nor would they care to understand it. Knowing she needed to speak the same language as the code she was reading, she took online genetics classes next.

"When the doctors brushed off my genetic data, I knew I had to take matters into my own hands," Julia recalls. "You have to say no to 'no.' That was the moment I knew the system wasn't going to hand me answers and that I had to find them myself." She wrote her own algorithms to analyze the data, comparing hypothetical combinations of her and her husband’s DNA against their sons’- DIY that ultimately led to the answer of the boys' true diagnosis: Aspartylglucosaminuria (AGU), an ultra-rare disease. She’d found the right backdoor. AGU may have just messed with the best and thanks to Julia, her sons might not be left to die like the rest.

She traveled to Finland, once a hub of AGU research, only to find progress stalled and labs deserted. There was that 'no' again. So she partnered with a French family and launched the Rare Trait Hope Fund and transformed it into a global call to action.

"We're not part of the 10%, we're the 0.1%, and that's our superpower," Julia says, boldly challenging the current narrative about rare disease being rampant. "Everyone keeps trying to make rare disease sound common to get attention, but maybe the real story is that we're rare for a reason. We're not diluted. We're focused. We're agile. We build tighter communities, we move faster, and we’re forced to innovate. That’s not a disadvantage - it’s an edge. Rare disease is like the internet was and it’s the next great breakthrough innovation." When I asked how she’s navigated the rare disease landscape she didn’t hesitate. She’s careful about who she lets in. "Trust," she says. "You test the signal. If something feels off - you cut the connection." She has built a trusted network carefully. No flashy allies, just a tight inner circle who share code, share risk, and have your back when the system crashes.

She has been busy since she was told "no." Julia developed a small molecule therapy - a "patch" published in Nature. It has helped stabilize the condition for her sons and other AGU patients while they work towards a more permanent system upgrade of a gene therapy. She's now preparing an IND application for the gene therapy, identifying the regulatory system's weak points and ready to execute her carefully planted code for ultimate success.

Low-key but relentless, Julia Taravella doesn’t wait for the system to catch up. In the world of rare disease, she isn’t Zero Cool or even 0.1 Cool - she’s Ultimate Cool. I have no doubt she’ll crack the code. And when she does, she’s going to share the data to help others. No spotlight, just impact. She may just close the laptop, disappear into the background and the system will be better because she touched it. Hack the planet and make a difference.

-By Effie Parks, CTNNB1 mom & storyteller at Once Upon a Gene. Bearing witness. Following starlight. Carrying the fuel.

Written for the @endrarediseases story series - real families, real stakes, and the science that just might catch up because of it.

[Startup Archive @StartupArchive_]

Michael Seibel on how to create a great startup pitch

In the clip below, former Y Combinator CEO Michael Seibel breaks down the two types of pitches every startup founder needs: a 30-second elevator pitch and a two-minute pitch for investors.

“A lot of people practice 10-minute, 30-minute, hour-long, pitches. I think that’s all garbage. I think you can get all of your points across in two minutes. And one thing I like to tell founders is that the more you talk, the more you have an opportunity to say something that people don’t like.”

30-Second Elevator Pitch:

You should be able to explain to anyone you come across what your company does in 30 seconds.

This should be three sentences:

1. What does your company do? Assume the person you’re talking to knows nothing. This should be a 1-sentence explanation that your mom or dad can understand (e.g. “We’re Airbnb and we allow you to rent out the extra room in your house” NOT “We’re Airbnb and we’re a marketplace for space”).

2. How big is the market? Do a couple hours of research so that you can give investors a rough approximation of the size of the market you’re in (e.g. Airbnb might give the size of online hotel booking market)

3. How much traction do you have? Ideally you can say something like: “We launched in January and we’re growing 30% month over month. We have $ X sales and Y users.” If you’re pre-launch, you need to convince investors that you’re moving quickly (e.g. “the team came together in January. By March we launched our beta. By April we launched our product.”).

Two-Minute Pitch:

This pitch is for people you’re actually trying to convince of something (e.g. investors, potential employees, etc.). You basically want to simply explain what you do and then ask for money.

There are 5 key components:

1. Clear 30 second pitch (everything mentioned above)

2. Unique insight—what do the biggest players in your market not understand? This should be 2 sentences.

3. How do you make money?

4. Team. If your team has done something that has made investors money, you should mention that (e.g. “we’re the founders of PayPal”). If you haven’t, don’t go on about the awards you’ve won or PhDs you hold. What investors want to hear is: how many founders? (hopefully 2-4) how many of the founders are technical? (hopefully 50% or more engineers) How long have you known each other? (ideally you’ve known each other either personally or professionally for at least 6 months) Are you all full-time?

5. The Big Ask ($$$). You have to know what you’re talking about when you ask for money. Are you raising on a convertible note or a SAFE? What’s the cap of the SAFE? How much money are you raising? What’s the minimum check size? If you don’t know these things, investors won’t think you’re serious or that you haven’t done your homework.

[Stefan Dvoretskii @StefanDvoretsk1]

Having fun feeding Grok Differential Geometry paper abstracts 👦

[Dan Goodman @neuralreckoning]

Another good reason to be on Mastodon. Check out Carl Bergstrom's absolute demolition of that preprint reported by Science that claims 30% of papers are fake and most from China. Brutal.

https://t.co/Zl8lKCv1FD

[Rafael Grigoryan @eegdude]

Hi!
I'm looking for an international research opportunity.
In the past, I did: a lot of BCI/EEG stuff both for academic and industrial applications (both startups and big tech)
I can: teach, do EEG/BCI research, write software, and prototype hardware.
https://t.co/jWesUv8Z2q

[Eric Topol @EricTopol]

The fundamental importance of basic science
https://t.co/C49sJr59uT
This week's @Nature editorial

[Stefan Dvoretskii @StefanDvoretsk1]

@matthen2 @readwise save thread

[Stefan Dvoretskii @StefanDvoretsk1]

"#Mastodon is [a] free open source software [analogon of Twitter,] and the #Fediverse [of which Mastodon is a part] cannot be bought" - quote @[email protected].

That's right! Catch me there too! @[email protected]

[Stefan Dvoretskii @StefanDvoretsk1]

@DanRosiak Sounds very interesting! Thank you for your effort and also Arcadia Fund for supporting Open Access 👏🏻

[Stefan Dvoretskii @StefanDvoretsk1]

As an example of a publication in this area, I post https://t.co/Q75NwKfuPo

[Stefan Dvoretskii @StefanDvoretsk1]

Besides, I'm too looking for a PhD position/Master's Thesis in Biological Computation. I would like to study brain organoids' signaling dynamics for that.
If you have heard of a lab doing this, please tell me!

[Stefan Dvoretskii @StefanDvoretsk1]

My specialization is more on the quantative side, though I have extensively worked with biologists/neurologists as a Research assistant. Both my B. Sc. and my M. Sc. are interdisciplinary (Bioinformatics and Neuroengineering).

[Luca M. Leisten | @lucaleisten.bsky.social @LucaLeisten]

Hi Twitter! Due to multiple reasons I decided to quit my PhD. I'm now looking for a (remote) research assistant/support/teaching position OR a new PhD position. Please let me know if you hear anything 👀 RTs appreciated! @PhDVoice #AcademicTwitter #jobsearch #phdchat #phdvoice

[Stefan Dvoretskii @StefanDvoretsk1]

@JesParent @ALifeConf @AlifePapers @balicea1 @Orthogonal_Lab Congrats back!

[Jes Parent 🧭 @JesParent]

Our @ALifeConf @AlifePapers Braitenberg Vehicles as Developmental Neursimulation paper has been fully published!

Congrats @StefanDvoretsk1, Ziki, Ankit, @balicea1, @Orthogonal_Lab !

[Stefan Dvoretskii @StefanDvoretsk1]

@CAPPO_GENGHIS Hello Cappo, when is your next concert?

[Clive Thompson @pomeranian99]

No matter how many times I mess around with Conway's Game of Life, I still find it magical when a pattern suddenly produces gliders

This pattern does it several times: https://t.co/CzpAtPd3mm

(Wraparound screen, so not really *canonical* Life, bien sur)

[Gerstner Lab @compneuro_epfl]

Interested in deep learning theory and its practical implications?
We are excited to host a virtual seminar at @EPFL_en on "Loss Landscape of Neural Networks" with a fantastic line-up of speakers.
Registration: https://t.co/4UK7zbtJa3