RFdiffusion

Search documents
诺奖得主David Baker最新Nature论文:AI设计蛋白开关,实现对药物的快速精准调控
生物世界· 2025-09-28 08:30
Core Viewpoint - David Baker's team has developed a groundbreaking AI protein design model, RFdiffusion, which allows for the precise control of protein-protein interactions, potentially revolutionizing fields such as cancer treatment and immune regulation [2][3]. Group 1: Research Breakthroughs - The new design method enables precise timing of cytokine signaling, allowing for "remote control" of protein interactions with second-level accuracy [3]. - The research focuses on designing the "excited state" of proteins, which influences the kinetics of protein-protein interactions, rather than just their stable states [7]. - A special "hinge protein" was designed to change conformation in response to external signaling molecules, facilitating rapid dissociation of protein complexes [10]. Group 2: Performance Metrics - The new design method achieves up to a 5700-fold increase in dissociation rates, allowing protein complexes that previously took hours to dissociate to do so in seconds [12]. - Structural analysis confirmed that the designed proteins closely matched theoretical predictions, with a maximum deviation of only 1.3Å [12]. Group 3: Applications - The technology has potential applications in developing rapid biosensors, such as a SARS-CoV-2 sensor with a response time of just 30 seconds, which is 70 times faster than previous sensors [14]. - It can create dynamic control circuits at the protein level, enabling efficient signal transmission and amplification [15]. - The method allows for the rapid shutdown of highly active splitting enzyme systems, providing new tools for metabolic engineering [16]. Group 4: Immunology Insights - The research has significant implications for controlling the interleukin-2 (IL-2) signaling pathway, which is crucial for immune response, allowing for rapid on/off switching of IL-2 analogs [18]. - Different durations of IL-2 stimulation were found to have distinct biological effects, with short stimulation providing anti-apoptotic protection while prolonged stimulation activated metabolic changes and cell division [19][20]. Group 5: Paradigm Shift in Protein Design - This research represents a paradigm shift in protein design, moving from static structure design to dynamic kinetic control, with broad applicability across various protein interactions [22]. - The technology not only serves as a powerful tool for basic biological research but also opens new avenues for therapeutic applications, potentially leading to more precise and controllable biotherapies [22].
诺奖得主David Baker推出RFdiffusion3,颠覆蛋白质设计格局,开启全原子生物分子设计新时代
生物世界· 2025-09-22 04:14
Core Viewpoint - The article discusses the advancements in protein design using generative artificial intelligence, particularly focusing on the breakthrough of RFdiffusion3, which allows for atomic-level precision in designing proteins that can interact with specific small molecules, DNA, and other biomolecules [9][24]. Group 1: RFdiffusion3 Overview - RFdiffusion3 represents a significant advancement in protein design, enabling the design of proteins with atomic-level precision, including interactions with non-protein components [9][10]. - The model is built on previous versions, RFdiffusion and RFdiffusion2, and offers improvements in accuracy, efficiency, and versatility [10][28]. - RFdiffusion3 can handle complex atomic constraints, such as hydrogen bonds and solvent accessibility, and is capable of designing various interactions, including protein-protein, protein-small molecule, and protein-nucleic acid interactions [10][28]. Group 2: Performance and Applications - In benchmark tests, RFdiffusion3 demonstrated superior performance with a computational cost only one-tenth of previous methods, making it significantly more efficient [3][10]. - The model has shown excellent results in designing DNA-binding proteins and enzymes, achieving a binding activity of 5.89±2.15 μM for a designed DNA-binding protein and a Kcat/Km value of 3557 for a designed cysteine hydrolase [21][28]. - RFdiffusion3 has outperformed its predecessor in multiple target designs, producing an average of 8.2 unique successful clusters compared to 1.4 from RFdiffusion [15]. Group 3: Technical Innovations - The core innovation of RFdiffusion3 lies in its all-atom diffusion model, which allows for simultaneous simulation of protein backbone and side chains, as well as interactions with non-protein components [9][10]. - The model employs a unified representation of amino acids, standardizing them to 14 atoms, which facilitates the handling of varying side chain atom counts [13][14]. - The architecture is based on a Transformer U-Net, which includes downsampling, sparse transformer modules, and upsampling to predict coordinate updates [14]. Group 4: Future Implications - The introduction of RFdiffusion3 marks a paradigm shift in protein design, enabling unprecedented control over complex functionalities, such as specifying enzyme active sites and controlling hydrogen bond states [24][25]. - As the technology continues to evolve, it is expected to lead to innovative therapies, new types of proteases, and biomaterials, fulfilling the vision of "designing life molecules" [25].
Cell重磅:AI从头设计生成小型结合蛋白,大幅提高先导编辑效率
生物世界· 2025-08-06 04:05
Core Viewpoint - The article discusses advancements in prime editing (PE) technology, particularly focusing on the development of MLH1 small binders (MLH1-SB) using AI tools to enhance editing efficiency in genome editing applications [2][4]. Group 1: Prime Editing Technology - Prime editing is a novel genome editing technique that allows for precise modifications, including base substitutions and small insertions or deletions [2]. - The efficiency of prime editing is often limited by the mismatch repair (MMR) pathway, which can hinder the integration of desired edits at target sites [6][7]. Group 2: AI-Driven Innovations - The research utilized the AI protein design tool RFdiffusion to create MLH1 small binders that inhibit MMR activity, thereby improving prime editing efficiency [3][9]. - AlphaFold3 was employed to efficiently screen the designed proteins, leading to the identification of an optimal MLH1-SB composed of only 82 amino acids, which integrates well with existing PE architectures [10][11]. Group 3: Efficiency Improvements - The newly developed PE-SB platforms, such as PEmax-SB, PE6-SB, and PE7-SB, demonstrated significant improvements in editing efficiency, with PE7-SB2 showing an increase of approximately 18.8 times compared to PEmax and 2.5 times compared to PE7 in human cells [11]. - In vivo studies indicated that PE7-SB2's efficiency was about 3.4 times greater than that of PE7 in mouse models [11]. Group 4: Implications for Gene Therapy - The compact size of the MLH1-SB allows for easier integration and delivery in gene therapy applications, which is crucial for effective in vivo gene editing [11]. - The advancements in AI-driven protein design are expected to facilitate the development of efficient gene editing therapies, potentially transforming the landscape of genetic medicine [15].
Nature/Science两连发:David Baker团队中国博后利用AI“驯服”无序蛋白,攻克“不可成药”靶点
生物世界· 2025-07-31 04:13
Core Viewpoint - Intrinsically disordered proteins (IDPs) and intrinsically disordered regions (IDRs) represent about 60% of the human proteome and are crucial for various cellular functions and disease progression. Recent advancements in artificial intelligence (AI) have enabled the design of specific binding agents for these previously considered "undruggable" targets, unlocking new therapeutic possibilities [1][2][20]. Group 1: Importance of IDPs and IDRs - IDPs and IDRs play significant roles in cellular signaling, stress responses, and disease progression, making them valuable targets for clinical diagnostics and drug development [2][8]. - Traditional drug design struggles with IDPs due to their lack of stable structure, which complicates the development of targeted therapies [6][7]. Group 2: AI Breakthroughs in Drug Design - The research led by David Baker's team utilized generative AI to design proteins that can accurately bind to IDPs and IDRs, achieving atomic-level precision [2][11]. - The AI model, RFdiffusion, allows for dynamic matching without pre-setting structures, enabling the generation of binding proteins that can adapt to the flexible nature of IDPs [11][12]. Group 3: Experimental Results and Applications - The studies published in Nature and Science demonstrated the successful design of binding proteins for various IDPs, with binding affinities ranging from 3 to 100 nanomolar [15][18]. - These binding proteins have shown potential in therapeutic applications, such as inhibiting amyloid fiber formation related to type 2 diabetes and disrupting stress granule formation in neurodegenerative diseases [16][18]. Group 4: Future Implications - The new design strategies developed could lead to innovative treatment methods and diagnostic tools for diseases associated with IDPs and IDRs, marking a significant advancement in precision medicine [20][24]. - The complementary strategies of RFdiffusion and logos provide a robust framework for targeting both structured and unstructured protein regions, enhancing the versatility of drug design [21][22].
攻克“不可成药”,David Baker团队中国博后利用AI从头设计蛋白,靶向内在无序蛋白,解锁治疗靶点
生物世界· 2025-07-19 03:06
Core Viewpoint - The article discusses the breakthrough in targeting intrinsically disordered proteins (IDPs) and intrinsically disordered regions (IDRs) using artificial intelligence (AI), specifically through the work of Professor David Baker and his team, which has made previously "undruggable" targets accessible for drug development [3][5][20]. Group 1: Research Breakthroughs - The research led by Professor David Baker utilizes generative AI to design proteins that can precisely bind to IDPs and IDRs, achieving atomic-level accuracy [3][5]. - The studies employed two complementary design strategies based on amino acid sequences, eliminating the need for structural information, thus enhancing the universality of drug discovery [7][22]. - The first study published in Science demonstrated the design of binding proteins for 43 diverse disordered protein targets, achieving tight binding for 39 of them, with affinities ranging from 100 picomolar to 100 nanomolar [14][20]. Group 2: Applications and Implications - The designed binding proteins show potential applications in various fields, including cancer treatment, disease diagnostics, and intervention in neurodegenerative diseases [14][20]. - Specific examples include a binding protein targeting enkephalin that successfully blocked pain signal transduction in human cells [14][21]. - The second study, available on bioRxiv, reported the design of binding proteins for various IDPs and IDRs, with affinities also in the range of 3-100 nanomolar [17][20]. Group 3: Methodology and Tools - The research utilized a protein design strategy called "logos," which created a library of binding pockets to recognize amino acid side chains, allowing for the assembly of binding proteins [9][11]. - The RFdiffusion model was employed to generate novel proteins that do not exist in nature, demonstrating its effectiveness in various therapeutic contexts [5][22]. - The strategies developed in these studies are now available online for researchers to use freely, promoting further exploration in the field [23][24].