Development of A Specialized Platform for Innovative Research Exploration (ASPIRE)

开发创新研究探索专业平台（ASPIRE）

• 批准号: 10908200
• 负责人: Samuel Michael
• 金额: $ 635.56万
• 依托单位: NATIONAL CENTER FOR ADVANCING TRANSLATIONAL SCIENCES
• 依托单位国家: 美国
• 资助国家: 美国
• 起止时间: 至
• 项目状态: 未结题
• 来源: https://reporter.nih.gov/project-details/10908200
• 关键词: Acceleration; Address; Adopted; Affect; Air; Artificial Intelligence; Artificial Intelligence platform; Automation; Award; Awareness; Back; Benchmarking; Binding; Biological; Biological Assay; Biological Process; Biology; Chemical Structure; Chemicals; Chemistry; Code; Collaborations; Communities; Computer Assisted; Coupled; Data; Data Scientist; Decentralization; Deposition; Descriptor; Development; Diagnosis; Disease; Electrospray Ionization; Emerging Technologies; Encapsulated; Engineering; Equipment and supply inventories; Extramural Activities; Face; Failure; Feedback; Government; Grant; Graph; Health; Human; Informatics; Infrastructure; Institution; Intelligence; Interdisciplinary Study; Intramural Research Program; Intuition; Ions; Laboratories; Laboratory Chemicals; Learning; Learning Module; Machine Learning; Maps; Mass Spectrum Analysis; Medicine; Methods; Microfluidics; Modeling; Modernization; Molecular; National Center for Advancing Translational Sciences; Nature; Organic Chemistry; Organic Synthesis; Pain; Patients; Pharmaceutical Preparations; Pharmacologic Substance; Pharmacology; Phase; Play; Positioning Attribute; Procedures; Process; Production; Professional Organizations; Programming Languages; Property; Protocols documentation; Qualitative Methods; Quantitative Structure-Activity Relationship; Reaction; Reagent; Recommendation; Reproducibility; Research; Research Personnel; Robot; Running; Sampling; Schedule; Science; Scientist; Secure; Site; Solvents; Spectrometry, Mass, Electrospray Ionization; Standardization; Structure; Surface; System; Systems Integration; Techniques; Technology; Testing; Translations; United States National Institutes of Health; Vision; Work; analytical method; blockchain; chemical reaction; chemical synthesis; cooking; cost; data integration; deep neural network; design; digital; digital twin; distributed ledger; drug discovery; effective therapy; experimental study; high throughput screening; improved; informatics tool; innovation; innovative technologies; instrument; interest; ionization technique; manufacture; mass spectrometer; nano; novel; novel strategies; novel therapeutics; open data; opioid use disorder; process optimization; programs; response; screening; simulation; technological innovation; timeline; vector; virtual

The nature of the ASPIRE program is such that the NCATS intramural program will not build it alone; rather it will be built in conjunction with extramural researchers who are leaders in the field of chemical and biological automation, engineering, informatics, and infrastructure. There were five (5) awards made in total; two (2) for physical hardware modules, and three (3) for virtual hardware modules. These modules will be developed in collaboration with NCATS during the UG3 period of the grant, and if successful, will be incorporated into the overall ASPIRE platform during the UH3 phase. Two Hardware Modules 1) U Glasgow (Lee Cronin) BioChemputer: An Intelligent Universal System for Running Automated Chemical Reactions Across Different Hardware and Scales NCATS Chemputer built using shared blueprints and testing plans underway, 2) Purdue U (Graham Cooks) High Throughput Infrastructure for Reaction Screening and Biology Factory and site acceptance testing completed for NCATS-bound platform using novel mass spec-based technique to synthesize and perform bioassays on compounds at nanogram scale. Three Virtual Modules 1) MIT (Connor Coley) Informatics and Machine Learning Modules for Reaction Planning, Scheduling, Simulation, and Optimization in the ASPIRE Autonomous Laboratory Using NCATS benchmark chemical reactions for optimization, 2) Purdue/IBRI/OnAI (Gaurav Chopra) Chemical Instruments-aware Distributed Blockchain Based Open AI Platform to Accelerate Drug Discovery NCATS BioRAPTR 2.0 dispenser in use as 1st real-world integration with Purdue AI platform, 3) Collaborative Drug Discovery (Barry Bunin) Virtual Approaches to New Chemistries NCATS chemists/data scientists testing platform for analysis, and designing integrated digital representations of molecules to navigate bioactivity and chemical reactivity. In this collaborative project we will develop a system for the automation and execution of chemical reactions across a range of hardware and scales for the synthesis of known and unknown molecules. This work will leverage the last 6 years of progress in our laboratory on the Chemputer (a programmable chemical synthesis robot) and the principle of Chemputation (the concept that a chemical synthesis expressed in a type of chemical code can be run on any compatible hardware reliably). Three specific aims are proposed: 1. To develop the chemical programming language standard that will run the synthesis protocols; 2. Development of modular plug and play hardware for chemical synthesis including interfaces to third party systems; and 3. Creation of a reaction screening system for chemical synthesis and discovery. Mass spectrometry (MS) is a powerful and widely applicable analytical method for qualitative and quantitative analysis of compounds of all types and sizes. Desorption electrospray ionization (DESI) is an ambient ionization method in which samples are analyzed in the open air by impact of primary droplets. Given the ability to position an array of samples relative to the mass spectrometer, DESI-MS becomes a high throughput (HT) chemical analysis method. The power of MS as an analytical method is well known but it is less commonly realized that MS is also a preparative method, e.g. it can be used to deposit mass-selected ions on surfaces to create new materials. A significant application from the point of view of organic synthesis, is that MS can be used to create microdroplets from reaction mixtures and, therefore, many reactions in confined volumes (especially in microdroplets) can be accelerated relative to their rates in bulk, often by orders of magnitude. A unique feature of DESI is that the spray of solvent used to analyze a reaction mixture generates secondary droplets upon impact and the reagents can react in the solution phase while the droplets are carried to the mass spectrometer. It is this remarkable feature that makes DESI-MS a powerful synthetic method combined with a built-in analytical capability. This collaborative project comprises the development of several virtual modules to support the multi-step chemical synthesis of new molecules in autonomous laboratories. These modules are designed to benefit traditional synthetic chemists in addition to automation chemists using the integrated hardware platform being developed by the ASPIRE team at NCATS. Computer-aided synthesis planning can be viewed as a hierarchical process of elaboration starting from the list of molecules of interest: (1) retrosynthetic planning to identify suitable starting materials and intermediates, (2) reaction condition recommendation to identify the conditions with which each reaction step should be run, (3) translation of hypothetical reaction steps into action sequences executable on automated hardware. Optional but valuable components include (4) recording procedures through an experimental planning module, (5) optimization of the timing and order of action sequences to most efficiently synthesize multiple synthetic targets via a digital twin of the platform, and (6) the iterative optimization of process parameters based on experimental responses in a feedback loop. Chemical instruments-aware distributed blockchain based open AI platform to accelerate drug discovery Artificial Intelligence (AI) and Automation has the potential to accelerate several stages of the drug discovery process, including the design-make-test-analyze optimization cycle, typically faced by medicinal chemists. However, several roadblocks exist resulting in too long timelines to deliver much needed innovation to patients with unmet needs. Both human and AI face similar limitations mainly due to disjointed steps needed to obtain and integrate the data that is generated by different organizations or laboratories and cannot be readily shared without disclosing IP sensitive information (e.g., non-patented novel chemical structures). In addition, there is lack of negative (failures) data available publicly, which are critical for generating accurate AI models, but are typically not made available outside of the originating institution or laboratory due to a variety of reasons related to IP. And, even among positive results, greater reproducibility of protocols is desirable. A solution to develop a fully integrated system in-house can be effective but it is hard to scale and not easily adopted mainly due to the costs and infrastructure involved. Our solution encapsulates the vision of NCATS ASPIRE program of integrating and automating laboratories to accelerate the drug discovery process while taking into account the above problems that exist. Blockchain, a distributed ledger technology, coupled with AI and Automation has the potential to solve all of the above problems as it has done in several other technology sectors, such as finance and medicine to securely share and learn from data without revealing its identity. We will develop a blockchain based open science AI framework as a decentralized laboratory cloud for the drug discovery community to enhance collaboration and reproducibility. Proposed as a modular component that will fit in with the large-scale automated synthesis program at NCATS and interoperate with other informatics tools such as retrosynthetic analysis and inventory management. The central technological innovation is our method for ttraining a deep neural network to do a graph-to-graph transformation. In the middle of the network is a chemically rich vector () which has two important properties: (1) it captures the structural diversity of the input molecules using a short vector of highly orthogonal values, making it highly effective for QSAR models without needing the additional descriptor-pruning step; and (2) vectors can be mapped back to the molecular space to recreate the original structure.