labriataphd's profile picture.

Luciano Abriata

@labriataphd

Check our new preprint, effort led by Stephen Buckley and @befcorreia, on the de novo design of phosphorylation-activated protein switches for synthetic signaling. In it, NMR unveiled how complex molecular mechanics actually make the designs work. 👉 biorxiv.org/content/10.110…

labriataphd's tweet image. Check our new preprint, effort led by Stephen Buckley and @befcorreia, on the de novo design of phosphorylation-activated protein switches for synthetic signaling.

In it, NMR unveiled how complex molecular mechanics actually make the designs work.

👉 biorxiv.org/content/10.110…

Luciano Abriata a reposté

🗓️ En el Día Internacional contra el Dengue, @GamarnikLab , jefa de nuestro Laboratorio de Virología Molecular, nos cuenta sobre el trabajo de su equipo que, desde hace 20 años, estudia a este virus que es una amenaza para la salud pública de la región.


Summary of a very cool idea

We’ve entered the age of AI-powered discovery. @labriataphd tells the story of a human-led team of GPT-4o agents that developed real, lab-validated nanobodies against COVID-19 and hints at what’s coming next. towardsdatascience.com/a-research-lab…



Luciano Abriata a reposté

We’ve entered the age of AI-powered discovery. @labriataphd tells the story of a human-led team of GPT-4o agents that developed real, lab-validated nanobodies against COVID-19 and hints at what’s coming next. towardsdatascience.com/a-research-lab…


My posts on the new Towards Data Science are all free to read. Check out this one about how Deepmind blended Genetic Algorithms with LLMs to evolve code and even new algorithms! towardsdatascience.com/googles-alphae…


All my articles are now free to read in the new Towards Data Science. Here's my latest!

Is @ChatGPTapp helping or hurting education? New meta-analysis reveals the ANSWER. @labriataphd meticulously analyzed 51 studies to quantify the TRUE impact of LLMs in educational settings. towardsdatascience.com/what-the-most-…



Luciano Abriata a reposté

I still can't believe all we accomplished with @labriataphd over these past 5 years. We truly made science fiction a reality and made a huge impact in democratizing access to digital learning tools, igniting interest in science and technology. The future is now!


I'm amazed at how well GPT-4o and even GPT-4o-mini can interpret figures, photographs, charts, screenshots even from molecular graphics software, etc. and extract information from them. Check out the several examples I tried, free to read here: towardsdatascience.com/testing-the-po…

towardsdatascience.com

Testing the Power of Multimodal AI Systems in Reading and Interpreting Photographs, Maps, Charts...

Can multimodal AI systems consisting in LLMs with vision capabilities understand figures and extract information from them?


Luciano Abriata a reposté

Of >105,000 participants with 30-year follow-up, only 9.3% achieved healthy aging (age 70, w/o any chronic diseases). Their diet was significantly associated with this outcome🧵 @NatureMedicine

EricTopol's tweet image. Of >105,000 participants with 30-year follow-up, only 9.3% achieved healthy aging (age 70, w/o any chronic diseases). Their diet was significantly associated with this outcome🧵 @NatureMedicine

Do you know of any work comparing Kds for protein-protein interactions obtained by ITC- and NMR? #nmrchat


Do you use structure prediction tools in your work? Then you must check this out. I just came across a review-like post summarizing all the results from CASP16; quite extensive and practical! nexco.ch/blog/Peeking-I…


What's the maximum # of acceptable 2FAs per day linked directly to work? Some 15-20 seems way excessive to me! We have things like AlphaFold and LLMs and nobody can come up with a better solution????


Still processing, or processing done???

labriataphd's tweet image. Still processing, or processing done???

Such cool insights! "[...] eventually, models will become [multimodal] enough [...] that they will connect different domains of expertise in novel ways, understanding genes not as mere letters but as molecular entities that can be targeted in terms [for example] drug discovery."

If you missed our blog last week: we covered how the rapid increase of data in Life Sciences allows scientists to train larger and better AI models. Those are now capable of predicting cell types, drug response and even genetic perturbations. Check it out: nexco.ch/blog/Foundatio…



Our article on molecularwebxr is out, showing how to use it for engaging discussions, outreach and education in immersive virtual and augmented reality that runs in all devices @epfl sciencedirect.com/science/articl…


Loading...

Something went wrong.


Something went wrong.