Private AI / RSI For Life Sciences & BioTech Labs
Earth
Joined December 2024
- Tweets494
- Following17
- Followers3K
- Likes807
LiteFold retweeted
Enterprises including small to mid scale biotechs should not face petty challenges on refusals. And thats where mixture of models works the best for this.
At LiteFold, we have pre-selected set of models inside our subagents. It' routes the request to the best model / subagent based on your request to come up with the perfect answer.
Links in the comment.
was the SARS-CoV-2 furin cleavage site bio-engineered??
i don't know, but the least you should have is an AI that doesn't refute the question to its lobotomized little brother. it's time to start answering some hard questions
@try_litefold
LiteFold retweeted
was the SARS-CoV-2 furin cleavage site bio-engineered??
i don't know, but the least you should have is an AI that doesn't refute the question to its lobotomized little brother. it's time to start answering some hard questions
@try_litefold
LiteFold retweeted
Great stuff, but why the model is comparing with protenix-v1 and boltz-1? When both of them are having v2 models.
Also, why are keep making co-folding models, with the same sequence + MSA + templates going to pairformer pairmixer backbone with all atom diffusion / flow matching head.
Yes, its going to work good, because we have oss data already available + more data collected from the lab. But there are so many other problems
- intrinsic water prediction
- cell-context-aware protein structure prediction.
- predicting structures which has very less homology like phages /viral proteins
- including glycans, PTMs, noncanonical residues
other things like speed, i mean yeah you can use tons of gpus for predicting 1000s of structure, but this solves for one time structure predictions. When you co-fold your backbones after inverse fold, there are similar problems, and all the hallucination based approaches on top of these models faces similar issues, where getting one trajectory takes so much amount of time.
I am still very bullish on how biohub started on ESM, I still have lots of hopes, and I personally consider this as a problem, I am very much excited to solve.
Introducing NovaAtom-Lite, an all-atom structure prediction model for drug discovery.
It predicts the structures of proteins, nucleic acids, and the molecules they bind, including how they fit together.
NovaAtom-Lite outperforms both IsoDDE and AlphaFold 3 on FoldBench antibody–antigen structure prediction (DockQ ≥ 0.23):
NovaAtom-Lite: 79.07% (n=172)
IsoDDE: 75.58%
AlphaFold 3: 47.90% (n=167)
Our evaluation uses 5 seeds × 5 samples, with the top-1 prediction selected by the model's own ranking score.
Better structure prediction. A stronger starting point for your next experiment.
Live today via Geodesic API and NovaDDE.
🔗 API: platform.geodesiclab.com/
🔗 NovaDDE: agent.geodesiclab.com
🔗 Model Card: geodesiclab.com/news/introdu…
LiteFold retweeted
I love how deep tech founders are using AI to transform biology
Biology is hard to generalize. But it is ready to specialize on your data and solve your problem.
Your lab already collects plenty of data. What it lacks are the systems that turn it into a learnable signal: state-of-the-art models, environments and evals, built around your problem.
LiteFold converts your own lab, into an AI learning environment, so you can recursively self improve for your the target you are working or the molecule you are designing.
We call this: Recursive Improvement From Experiment Feedback (RIEF)
Try LiteFold today and contact us if you want to integrate LiteFold to your lab.
LiteFold retweeted
Now I know, there will be a lot of questions around this. How much needle this is gonna move. To answer all of that, I will post a detailed blog on this soon!
So Stay tuned!
Biology is hard to generalize. But it is ready to specialize on your data and solve your problem.
Your lab already collects plenty of data. What it lacks are the systems that turn it into a learnable signal: state-of-the-art models, environments and evals, built around your problem.
LiteFold converts your own lab, into an AI learning environment, so you can recursively self improve for your the target you are working or the molecule you are designing.
We call this: Recursive Improvement From Experiment Feedback (RIEF)
Try LiteFold today and contact us if you want to integrate LiteFold to your lab.
LiteFold retweeted
A lab gets better when the last experiment changes the next design.
Using LiteFold, we ran a peptide campaign against BHRF1. We got 47% in hit rate.
That measurement is the whole point of LiteFold: your data, your target, the next round smarter than the last.
Lots of blogs and technical reports are on the way!
Biology is hard to generalize. But it is ready to specialize on your data and solve your problem.
Your lab already collects plenty of data. What it lacks are the systems that turn it into a learnable signal: state-of-the-art models, environments and evals, built around your problem.
LiteFold converts your own lab, into an AI learning environment, so you can recursively self improve for your the target you are working or the molecule you are designing.
We call this: Recursive Improvement From Experiment Feedback (RIEF)
Try LiteFold today and contact us if you want to integrate LiteFold to your lab.
LiteFold retweeted
Biology is hard to generalize. But it is ready to specialize on your data and solve your problem.
Your lab already collects plenty of data. What it lacks are the systems that turn it into a learnable signal: state-of-the-art models, environments and evals, built around your problem.
LiteFold converts your own lab, into an AI learning environment, so you can recursively self improve for your the target you are working or the molecule you are designing.
We call this: Recursive Improvement From Experiment Feedback (RIEF)
Try LiteFold today and contact us if you want to integrate LiteFold to your lab.
LiteFold retweeted
Turning manufacturing data into focused investigations with LiteFold.
A manufacturing process rarely loses efficiency all at once. Sometimes, the warning appears one batch at a time.
In LiteFold’s Dupilumab manufacturing example, recovery during Protein A purification fell from 92% to 75% across three consecutive batches. The numbers revealed a problem. Finding its source required looking across the process.
Examining manufacturing data and ranking potential causes to build a focused investigation plan, Targeted studies and a corrective action strategy gave the team a clearer path from observation to investigation.
litefold.ai
RT @try_litefold: A manufacturing process rarely loses efficiency all at once. Sometimes, the warning appears one batch at a time.
In Lite…
LiteFold retweeted
You can replicate such type of experiments on our platform. You can also use our API and SDKs, to just use the compute and custom pipelines or even agent to agent controls.
DM's open for early api access.
LiteFold retweeted
I did not even realize that during this 1hr of hill climb we predicted 600+ structures (6 models x 100 complexes), 6 campaigns all in parallel.
And now you just connect with your wet lab and wait for the experiments to finish and then update all the required components and climb higher.
