Hongshan Li’s Post

#Alphafold3 Proteins are long chains of amino-acid residues that fold into specific shapes. Properly folded proteins function normally whereas misfolded ones can lead to debilitating diseases. Google subsidiary named DeepMind developed a purpose-built AI tool to predict the shapes into which different proteins could fold, called AlphaFold in 2018.The upgraded AlphaFold 2 followed two years later. Recently, DeepMind launched AlphaFold 3, which can reportedly predict the shapes with nearly 80% accuracy as well as model DNA, RNA, ligands, and modifications to them. As with the first two AlphaFolds, no. 3 is great for being able to elucidate the folded proteins’ structures in seconds rather than the years humans have required with advanced microscopic techniques. These machines can predict protein structures with relatively high accuracy but they cannot say why they are folded that way; this is still the task of human scientists.

𝙊𝙣𝙚 𝙤𝙛 𝙩𝙝𝙚 𝙢𝙤𝙨𝙩 𝙞𝙢𝙥𝙤𝙧𝙩𝙖𝙣𝙩 𝘼𝙄 𝙢𝙤𝙙𝙚𝙡𝙨 𝙞𝙣 𝙗𝙞𝙤𝙡𝙤𝙜𝙮 𝙟𝙪𝙨𝙩 𝙜𝙤𝙩 𝙖 𝙢𝙖𝙟𝙤𝙧 𝙪𝙥𝙜𝙧𝙖𝙙𝙚. Google DeepMind and Isomorphic Labs debuted #AlphaFold3 on Wednesday, which includes a dramatic expansion of the model's capabilities to generate three-dimensional structures of biology. The previous version, #AlphaFold2 (released 2021), focused on taking an amino acid sequence of a protein and quickly generating that protein's 3D structure. The newer version, which incorporates a diffusion model into the mix, can generate far more complex structures that include proteins, DNA, RNA, small molecules, and more biomolecules. Isomorphic’s chief AI officer Max Jaderberg had this to say. “That’s really powerful for understanding more and more about the context of these molecules in the cell”, “AlphaFold 3 expands on the transformative advance that was AlphaFold" and “In short, it does it bigger, faster, better.” Read Andrew Dunn latest edition on Endpoints News for more news about AlphaFold.

Last week, I attended the #spatialomics24 conference in the beautiful city of Ghent. While I was highly impressed by the intellectual effort and hard work invested in developing new technologies, analysis pipelines, and AI tools, I started to wonder about the field’s ultimate contribution to discovering new biology. Undoubtedly, there were some novel findings, but they often came at a high cost and could, in the end, be recapitulated by simple antibody staining. For example, different zones in tumors discovered by multi-omics techniques were already visible on the H&E slide. High-dimensional omics datasets were reduced back to basic cell types (what are hundreds of transcripts worth if at the end I just use CD8 expression?). AI was used to generate images from UMAPs (but to what end?). Don't get me wrong, the field is still in its early stages, and promising technologies are emerging daily. I was particularly impressed by label-free tools, including Raman imaging. However, we might need to refocus on the biology itself more frequently. We should start testing hypotheses in real experiments and re-examining what constitutes correlation versus causality. And this might mean to include reporter and recording systems into the OMICs pipeline such as LIPSTIC (labelling immune partnerships by SorTagging intercellular contacts), fate-mapping/barcoding with Polylox/Polytope or timers such as Zman-seq.

"A combined team of medical researchers and AI systems specialists from Google's Deep Mind project and Isomorphic Labs, both in London, has made what the group describes as substantial improvements to AlphaFold 2 that make it possible for the application to predict the structure of a wide variety of biomolecular systems more broadly and accurately. The new iteration is called AlphaFold 3." #alphafold3

Introducing AlphaFold 3, a new AI model developed by Google DeepMind and Isomorphic Labs. By accurately predicting the structure of proteins, DNA, RNA, ligands and more, and how they interact, we hope it will transform our understanding of the biological world and drug discovery. https://lnkd.in/dgsr2fsC #googleai #ai #deepmind

🚀 AlphaFold 3: A New Era in Biological Understanding and Drug Discovery Google DeepMind and Isomorphic Labs have introduced AlphaFold 3, a groundbreaking AI model that predicts the structure and interactions of life's molecules with unprecedented accuracy. 🌍 ✨ Key Highlights: ❐ 50% improvement in predicting protein interactions compared to existing methods. ❐ Predicts interactions between proteins, DNA, RNA, ligands, and more. ❐ Free access through the newly launched AlphaFold Server, empowering researchers globally. AlphaFold 3 is set to revolutionize our understanding of the biological world and accelerate drug discovery. #AI #AlphaFold3 #DrugDiscovery #GoogleDeepMind #IsomorphicLabs

This will change everything! AlphaFold3 is going to significantly add value to early-stage drug discovery. When I started my venture, I was told numerous times how little computation is used in drug discovery. However, as a builder, I take this as an indication that there are many things that can be improved, and there is a bigger future for computational drug discovery. I feel lucky to be in an era where AlphaFold3 and other AI technologies are rapidly evolving. The advancements in AI X biology could potentially reshape our understanding of life and lead to groundbreaking innovations that benefit humanity for generations to come. This model will have its place in the annals of in human civilization. There is still a lot of hope and brightness in this world. #AF3 #drugdesign

Exciting news for the field of AI! Google DeepMind has released an improved version of its biology prediction tool, AlphaFold. AlphaFold 3 can predict the structures of DNA, RNA, and proteins, as well as other essential molecules for drug discovery, with more accuracy than ever before. According to DeepMind, this tool provides a more nuanced and dynamic portrait of molecule interactions than anything previously available. Check out the article for more information on this real, practical use case for advancement in the field of AI. https://lnkd.in/gwMYkVga

🔬 Google DeepMind's AlphaFold 3 Revolutionizes Molecular Modeling 🔬 Google DeepMind's latest breakthrough, AlphaFold 3, extends its capabilities beyond proteins to DNA, RNA, and other molecular structures. This leap in precision enhances research across medicine, agriculture, and more, showcasing a remarkable 50% improvement in prediction accuracy! 🧬 With applications in drug discovery and a commitment to responsible deployment, AlphaFold 3 is setting new standards in the use of AI for scientific advancement. #DeepMind #AlphaFold3 #ArtificialIntelligence #ScienceInnovation

Discovering a new target or molecule can take years, and involves predicting protein functions, identifying metabolic pathways, and analyzing protein interactions. To accelerate drug discovery, we now use deep learning-based supervised algorithms that create 3D models using amino-acid sequences. Recently, Novartis and Lily signed a deal with Isomorphic platform by Deepmind, which is built on the AlphaFold, an AI-based protein structure repository. It will be exciting to see how these developments will help accelerate drug discovery. #ai #drugdiscovery