MARA, or Molecular Analysis and Reasoning Assistant, represents the latest generation of LLM-enabled applications, functioning as an AI Agentic system specifically tailored for Scientific Informatics. This scientific co-pilot operates entirely behind an organization's IT firewall, ensuring data privacy and security. It empowers scientific teams, particularly in Cheminformatics and Bioinformatics, to effortlessly construct and integrate custom tools and complex workflows. MARA seamlessly connects with internal databases and existing tasks, streamlining research processes. For Cheminformatics, MARA aids in industry-grade drug design, integrating with scientific workflows and facilitating global collaboration. In Bioinformatics, MARA acts as a central hub, coordinating and distributing genomic and proteomic tools to researchers, allowing bioinformatic workflows to adapt naturally to evolving needs by integrating solutions spread across the organization. The system provides clear visibility into tool usage and input/output, mitigating AI hallucinations and fostering trust in results. It transforms Python snippets and API endpoints into accessible infrastructure, making them instantly available across the organization as managed, sandboxed services. MARA also features a natural language data engine for curating datasets and manipulating data frames, and an interactive molecular viewer for visualizing and sketching 2D molecules. Its robust workflow creation capabilities allow for defining and executing multi-step processes, enhancing data analysis and collaborative research, and can be prepared in MARA for collaborative analysis in Nanome XR.

Features & Benefits

• Scientific AI Co-pilot
  • An AI Agentic system for Scientific Informatics, leveraging LLMs to act as a co-pilot for scientific discovery.
• Secure & Private Deployment
  • Operates completely behind your IT firewall, supporting preferred LLMs like LLAMA 3 or privately Azure-hosted GPT4, with SSO-based login and key usage monitoring. Nanome has no access to enterprise data.
• Tool Creation & Integration
  • Enables easy building or integration of existing tools using a variety of methods, making them instantly available organization-wide.
  • Make Python Snippets accessible infrastructure: Create tools from Python code or AI-generated code, automatically creating managed, sandboxed services and REST Endpoints.
  • Access 100+ Informatics Tools: Utilize an extensive list of built-in tools.
  • API Endpoints: Quickly create tools to interact with API endpoints for existing infrastructure or external services.
  • Query Databases: Make typical SQL queries accessible via reusable templates, generate SQL on the fly, or construct queries using natural language.
• Natural Language Interaction
  • Facilitates scientific tasks through intuitive natural language prompts.
  • Visualize Molecules: Manipulate atomic representation of molecular structures and automatically embed interactive molecular viewers in chat.
  • Quickly Sketch 2D Molecules: Built-in 2D chemical editor for rapid ideation and drawing.
  • Natural Language Data Engine: Automatically curates datasets from chat, manipulates data frames, and expands datasets using plain English.
• Custom Workflow Creation
  • Define and execute complex, multi-step informatics workflows.
  • Specify consecutive tools and execute workflows.
  • Example workflow steps: Retrieve PDB Codes, Extract Ligands, Fetch Similar Molecules, Calculate Chemical Descriptors.
  • Reuse and distribute workflows company-wide.
• Domain-Specific Support
  • Tailored functionalities for key scientific disciplines.
  • For Cheminformatics: Aids in industry-grade drug design, integrates with scientific workflows, and supports global collaboration.
  • For Bioinformatics: Coordinates and manages distribution of genomic and proteomic tools, adapting workflows to research needs.
• Trustworthy Results
  • Provides scientists with clear visibility into tool usage, input, and output, actively mitigating AI hallucinations.