basedlabs

Open source virtual biolab

[basedlabs @based_labs]

Big news! We’ve teamed up with @adaptyvbio to take protein design to the next level.

We design EGFR binders using RFDiffusion, ProteinMPNN, and ESMFold—and then use Adaptyv Bio API to send the designs straight to their lab for testing. 🧪🔥

From computer to the lab in one smooth workflow.

[basedlabs @based_labs]

We all know that researchers spend a significant amount of time searching arXiv for relevant papers. In our spare time, we decided to tackle this problem head-on.

Introducing AI-powered arXiv Search 🔥 🔍 : A tool that can quickly pinpoint and summarize research topics from arXiv, cutting down the time spent sifting through countless papers.

After spending 14 hours vectorizing abstracts from arXiv papers up to 01/12/2024, we integrated them into a Retrieval-Augmented Generation (RAG) system. We now leverage PaperQA as the RAG engine, improving upon the traditional in-memory approach by utilizing a dedicated vector database of abstracts. It’s fully open-source — including the vector DB.

Check it out and let us know what you think! https://t.co/3nP1ZysEaL

[basedlabs @based_labs]

NoLabs tutorial 1: Bioinformatics made easy

[basedlabs @based_labs]

We believe that many new and interesting discoveries will be done with existing and emerging bio computational tools, so if you are interested in using them but don't really wanna code - check out NoLabs

[basedlabs @based_labs]

WE ARE BACK.

We've been silent for some time because we were cooking something really special, as we promised.

We have built the scalable platform for running in silico experiments, with no boundaries.

[basedlabs @based_labs]

We are building a platform that provides a user-friendly interface, APIs, and an AI copilot agent to streamline work with:

-Biological data access
-Protein folding models
-Protein solubility and function prediction
-Binding site prediction
-Protein-ligand docking
-De-novo molecule design
-Single-cell gene expression prediction
-Various other bioinformatics tools

[basedlabs @based_labs]

Generate small molecules for any protein target

[basedlabs @based_labs]

@lewischewis 3TBs actually

[basedlabs @based_labs]

@elisagdelope Big things coming soon😎

[basedlabs @based_labs]

Soon you’ll be able to benchmark any model supported by NoLabs on the data you want by just a simple query, run any model from a chat interface, and even ask about the results (“How well do you think a certain drug would perform for this disease?” etc.).

Exciting things are coming, and we’d appreciate any ideas or contributions. NoLabs is totally open source: https://t.co/GN9M3UNg6s

[basedlabs @based_labs]

BioBuddy-1: open source drug discovery co-pilot

[basedlabs @based_labs]

We’ve been adding different bio-AI models to NoLabs, but we quickly ran into an issue: it’s super hard to test them, and even harder to think about what data we can use for them and how to get it. So, we’ve decided to build our own thing - BioBuddy-1.

BioBuddy can query ChemBL and RcsbPDB to download any drugs, molecules, and proteins which are interesting for your research. For example, to test how a certain model performs on some protein family, such as rhodopsins, you can ask it, “Can you pull me all rhodopsins?” And it will download them for you.

The same goes for drugs - “Can you pull me 100 drugs which are going through clinical trials right now?”, “What does an ibuprofen drug look like?”, “Can you download 30 drugs related to lymphoma?” etc., are all valuable questions which you can use to do so.

And, of course, you can ask any question about drug research, chemistry, or biology.

This speeds up the research at least 10X for us and hopefully for other people, but we are only starting to release what we’ve been working on.

[basedlabs @based_labs]

DiffDock is now supported on NoLabs: https://t.co/RqL7m8vrK7

Access it easily via APIs or using our Docking workflow, straight from your browser

[basedlabs @based_labs]

After rewriting NoLabs virtually from the ground up, we are excited about the acceleration we are about to experience.

For the last several weeks, we've been revising our infrastructure to make it scalable and easy to use. Now, we're all set to rapidly expand with new models and drug discovery features.

NoLabs now supports:

- Microservices for each model
- APIs
- A completely revised UI

We are also SUPER excited to announce that in the coming weeks, we will produce our first report on Ewing's sarcoma. We will start with known targets and design drugs to combat this disease. Completely with No Labs.

[basedlabs @based_labs]

Based January

[Journal of Cheminformatics @jcheminf]

new: "Structure-based, deep-learning models for protein-ligand binding affinity prediction" https://t.co/mP55wysq4p

[basedlabs @based_labs]

Thanks for shouting us out🫡 Yes, it's still early days for us, so we are working hard right now to make it truly valuable and insightful. The main idea is to make a flexible infrastructure, so when the new docking/folding/inference/generative models come to market we can quickly add them and make available for everyone in "production" mode.

Also we are not just moving models to production like neurosnap, we try to squeeze out as much as we can from ensemble of different models to provide many more insights than just UI of inputs/outputs for a single model.

If you have any ideas of what would be cool to add feel free to reach out!

[basedlabs @based_labs]

Gonna start our own drug discovery research this year with our own infrastructure to push the boundaries and put some skin in the game

[basedlabs @based_labs]

If you are a small lab struggling to have your own infrastructure with state of the art inference models for drug discovery or protein research we are here to help (let's chat)