Medvolt’s Post

Excited to share some groundbreaking insights on leveraging DNA-Encoded Libraries (DEL) and focal molography to accelerate drug discovery! 🚀 Get the document here: https://lnkd.in/epvFvG_T 🔬 What’s the breakthrough? DEL technology has revolutionized how we screen millions of compounds for potential drugs, using unique DNA barcodes. But the real game-changer? Focal molography. This innovative technique allows us to: ✅ Validate up to 54 hits in a single experiment ✅ Analyze binding kinetics with unmatched precision ✅ Work efficiently with real biological samples 🎯 Why this matters In a field where every minute counts, this integration is a quantum leap for discovering therapies faster and more efficiently. It’s about turning potential into proof—quickly and reliably. 💡 Curious about the science? Here's a snapshot: DEL compounds are immobilized via DNA barcodes. With focal molography, interactions like binding affinity (KD) are quantified with minimal noise, thanks to the unique "mologram" sensors. 💬 Now over to you: What’s your take on the future of drug discovery with these innovations? Are you exploring DELs or similar tech? Let’s spark a conversation! #DrugDiscovery #BiotechInnovation #FocalMolography #DNAEncodedLibraries #BiomedicalResearch Let me know how it resonates, and feel free to share your perspectives below! 🌟

Molecular acumen, molecular synthetic data, generative chemistry, biomolecular simulations, deep chemical universe, drug design...add these concepts to your vocabulary. You will hear more and more of InVirtuoLabs, an AI-driven drug discovery and design biotech. https://lnkd.in/dtUgmXNy

𝗣𝗼𝗰𝗸𝗲𝘁𝗚𝗲𝗻: 𝗔 𝗕𝗿𝗲𝗮𝗸𝘁𝗵𝗿𝗼𝘂𝗴𝗵 𝗶𝗻 𝗣𝗿𝗼𝘁𝗲𝗶𝗻 𝗣𝗼𝗰𝗸𝗲𝘁 𝗗𝗲𝘀𝗶𝗴𝗻 Researchers Zaixi Zhang, Wanxiang Shen, Qi Liu, and Marinka Zitnik have developed PocketGen, a cutting-edge deep generative method for creating high-fidelity protein pockets that bind ligand molecules with exceptional precision. 𝗪𝗵𝗮𝘁 𝗺𝗮𝗸𝗲𝘀 𝗣𝗼𝗰𝗸𝗲𝘁𝗚𝗲𝗻 𝘀𝘁𝗮𝗻𝗱 𝗼𝘂𝘁? 1️⃣ 𝗗𝘂𝗮𝗹 𝗠𝗼𝗱𝘂𝗹𝗲 𝗦𝘆𝘀𝘁𝗲𝗺: - Bilevel Graph Transformer: Captures multi-granularity interactions at the atom and residue/ligand levels - Sequence Refinement Module: Utilizes protein language models (pLMs) with structural adapters for sequence prediction, ensuring sequence-structure consistency 2️⃣ 𝗘𝗳𝗳𝗶𝗰𝗶𝗲𝗻𝗰𝘆 𝗮𝗻𝗱 𝗣𝗲𝗿𝗳𝗼𝗿𝗺𝗮𝗻𝗰𝗲: - Trained on a vast dataset, including over 200 million predicted protein structures from the AlphaFold database - Outperforms existing methods, achieving a 95% success rate in generating pockets with higher binding affinity than reference pockets 3️⃣ 𝗩𝗲𝗿𝘀𝗮𝘁𝗶𝗹𝗲 𝗔𝗽𝗽𝗹𝗶𝗰𝗮𝘁𝗶𝗼𝗻𝘀: - Can be used to design enzymes, biosensors, and therapeutic proteins with customized properties - Effective in generating diverse and novel protein pockets for various ligands, enhancing the scope of protein engineering and drug discovery 𝗞𝗲𝘆 𝗙𝗲𝗮𝘁𝘂𝗿𝗲𝘀: - 𝗥𝗮𝗽𝗶𝗱 𝗚𝗲𝗻𝗲𝗿𝗮𝘁𝗶𝗼𝗻: 10x faster than traditional physics-based methods - 𝗛𝗶𝗴𝗵 𝗕𝗶𝗻𝗱𝗶𝗻𝗴 𝗔𝗳𝗳𝗶𝗻𝗶𝘁𝘆: Achieves superior Vina scores and designability metrics compared to state-of-the-art methods like RFDiffusionAA - 𝗦𝗲𝗾𝘂𝗲𝗻𝗰𝗲-𝗦𝘁𝗿𝘂𝗰𝘁𝘂𝗿𝗲 𝗖𝗼𝗻𝘀𝗶𝘀𝘁𝗲𝗻𝗰𝘆: Incorporates evolutionary information for accurate sequence prediction PocketGen opens new frontiers in the design of small-molecule-binding proteins, offering unprecedented accuracy and efficiency. 𝗛𝗲𝗿𝗲 𝗮𝗿𝗲 𝘁𝗵𝗲 𝗶𝗺𝗽𝗼𝗿𝘁𝗮𝗻𝘁 𝗿𝗲𝗳𝗲𝗿𝗲𝗻𝗰𝗲 𝗹𝗶𝗻𝗸𝘀: 📜 Paper: https://lnkd.in/dZ4gV9rP 𝐅𝐞𝐞𝐥 𝐟𝐫𝐞𝐞 𝐭𝐨 𝐜𝐨𝐧𝐭𝐚𝐜𝐭 𝐮𝐬 𝐢𝐟 𝐲𝐨𝐮 𝐡𝐚𝐯𝐞 𝐚𝐧𝐲 𝐢𝐧𝐪𝐮𝐢𝐫𝐢𝐞𝐬 𝐨𝐫 𝐫𝐞𝐪𝐮𝐢𝐫𝐞 𝐚 𝐝𝐞𝐦𝐨𝐧𝐬𝐭𝐫𝐚𝐭𝐢𝐨𝐧. 𝐖𝐞'𝐫𝐞 𝐡𝐞𝐫𝐞 𝐭𝐨 𝐚𝐬𝐬𝐢𝐬𝐭 𝐲𝐨𝐮 𝐚𝐧𝐝 𝐩𝐫𝐨𝐯𝐢𝐝𝐞 𝐜𝐥𝐚𝐫𝐢𝐭𝐲. Visit our website: https://www.medvolt.ai or reach out to us via email: contact@medvolt.ai Follow Medvolt for more such updates #drugdiscovery #proteinengineering #bioinformatics #machinelearning #AI

Our team just launched a 765M parameter molecule generation model, fine-tuned on a select subset of our custom LS dataset, now available on Hugging Face (closed access, by request only). This model enables molecule generation with targeted properties, helping to shape the future of molecular AI. 👉 Interested? Connect with our support team for access! #DrugDiscovery #AI #MoleculeGeneration #HuggingFace #CollAIborate #ContactDoctor

Did you know that X-Chem is a contributor to the The Structural Genomics Consortium (SGC)'s #Target2035 initiative, an open science global movement focusing on the creation of chemical and biological tools to study human proteins and inform drug discovery? As a pioneer of DNA-encoded chemical library technology, we are generating quality datasets from DEL screens of individual targets, curating the datasets so that they are compatible with machine learning, and through the SGC, posting the data to the public portal “AIRCHECK” for researchers around the world to access and build models with. Target2035 aims to develop a pharmacological modulator for every protein in the human proteome by 2035. While still in its early phase, AIRCHECK is now available: https://aircheck.ai/ #DataScience #MachineLearning #Sustainability #OpenAccess

🚀 We are pleased to announce a significant development for the Chemistry and Drug Discovery community. 🚀 Our new Molecule Generation Model is now available on Hugging Face. This model, featuring 765 million parameters and fine-tuned on a smaller subset of our custom Life Sciences (LS) dataset with ChEMBL insights, allows users to generate molecules tailored to specific molecular properties, including weight, LogP, functional groups, as well as toxicity and stability considerations. This launch is part of our commitment to advancing molecular sciences through the fine-tuning of smaller, efficient models that incorporate specialized domain knowledge. Please be advised that this model is currently closed and can be accessed upon request. We encourage those interested in utilizing this technology to contact our support team for access and to provide feedback regarding its functionality and potential applications. 👉 To request access, please reach out to our support team: info@collaiborate.com Let us collaborate to drive innovation and shape the future of molecular artificial intelligence. #AI #DrugDiscovery #Chemoinformatics #MachineLearning #HuggingFace #MoleculeGeneration #Innovation #CollAIborate #ContactDoctor #Lifesciences

How can Machine Learning (ML) increase the production of medicine? 💊 And how far is the pharma industry with adopting Industry 4.0 technologies such as ML? Watch the video below 📹 - and learn about the answers in this article: https://lnkd.in/dPa_mb7R In the video, MADE PhD from DTU - Technical University of Denmark Breno Renato Strüssmann Junior tells about his research on applying ML and other I4.O technologies at FUJIFILM Diosynth Biotechnologies, Hillerød. The research is part of the MADE FAST research platform including more than 50 company partners, universities and RTOs. The largerst investment in the platform comes from Innovationsfonden. Partners in the research project: DTU, FORCE Technology and FUJIFILM Diosynth Biotechnologies. #Industry4 #MachineLearning #DKpharma #Hillerød #dkproduktion

🚀 Excited to announce our latest work, #InstructMol, which has been accepted at #NeurIPS2024! 🎉 🌟 Tackling Data Scarcity in Biochemistry with Semi-Supervised Learning An enduring challenge in biochemistry and molecular modeling is the data scarcity problem. Unlike other fields where labeled data is abundant, generating biochemical labels requires costly and time-consuming experiments, which limits the potential of machine learning (ML) in drug discovery and molecular property prediction. 💡 Enter InstructMol—our novel approach to unlocking the potential of unlabeled data for robust molecular property prediction. 🔑 Key Highlights: 1️⃣ Instructor Model: InstructMol introduces an instructor model to estimate the confidence of predicted labels, generating reliable pseudo-labels for unannotated data. 2️⃣ Seamless Utilization of Unlabeled Data: Our method avoids the typical pitfalls of domain transfer by enabling the molecular model to learn from pseudo-labeled data, eliminating the pretrain-finetune gap directly. 📊 Impact: State-of-the-art performance: Achieved on MoleculeNet and several OOD benchmarks. Real-world application: Successfully predicted properties of 9 newly patented drug molecules (ZA202303678A), demonstrating InstructMol's potential in real-world drug discovery scenarios. Feel free to reach out if you’d like to learn more about our work or discuss potential collaborations! #PaperLink: https://lnkd.in/g-h5gNQj #Code: https://lnkd.in/gihMFvED #NeurIPS2024 #AIForScience #DrugDiscovery #MachineLearning

Five years of lab work, hundreds of experiments, and now sit on a goldmine of untapped data - ready to be transformed. The Nobel Prize winners in Chemistry 2024 showcased the power of AI in molecular science, sparking your curiosity, but you don't know where to begin. At Ingenie Bio, we bridge the gap between raw data and research discoveries so you don't have to navigate AI complexities alone. Our founder, Fabien Plisson, Ph.D., has spent countless hours in the lab synthesizing analogs from natural products, small molecules, and peptides. By mastering statistics, data science, and chemo/bioinformatics, he turned complex datasets into insights, leading to potent kinase inhibitors and GPCR peptide ligands. How much time could you save with our expert services? If you are ready to turn your data into biomolecular discoveries, drop 'START' below or contact us at https://lnkd.in/gpg2ruQB #AI #MachineLearning #ProteinDesign #DrugDiscovery #Biotech

Strawberry Creek Ventures' portfolio company, Menten AI, is a biotechnology company with a generative AI platform for design and optimization of peptide macrocycles.    This week, the company announced the completion of a previously undisclosed research collaboration and licensing agreement with Bristol Myers Squibb, utilizing Menten AI’s platform to optimize cyclic peptides. Menten AI’s platform integrates advanced machine learning techniques with physics-based models and quantum chemistry simulations to expedite the design and optimization of peptide macrocycles. Instead of screening millions of molecules, Menten AI uses generative AI to efficiently navigate the chemical space and identify macrocycles with the desired properties. This approach significantly reduces the number of candidate molecules needing wet lab testing and the overall iterations required to achieve drug-like properties. #Cal #UCBerkeley