Sparks: Multi-Agent Artificial Intelligence Model Discovers Protein Design Principles

TL;DR

Sparks introduces a multi-modal, multi-agent framework that autonomously conducts the full cycle of scientific inquiry—hypothesis generation, experiment design, execution, and reporting—without human input. By coupling generation and reflection across specialized agents, Sparks explores protein design problems beyond its training distribution, uncovering a length-dependent mechanical crossover in peptides ($L$ crossing $80$ residues) and a chain-length/secondary-structure stability landscape that highlights robust $eta$-sheet–rich designs and a pronounced ‘frustration zone’ for mixed folds. The work demonstrates end-to-end autonomous discovery, validated through two in-depth protein-discovery case studies, and discusses limitations and future directions toward domain-general AI laboratories. The approach leverages de novo protein design (Chroma), folding (OmegaFold), unfolding-force prediction (ProteinForceGPT), and physics-based simulations (NAMD with CHARMM/GB), and establishes a benchmark for out-of-distribution creative science powered by a tightly integrated generation–reflection loop with structured documentation.

Abstract

Advances in artificial intelligence (AI) promise autonomous discovery, yet most systems still resurface knowledge latent in their training data. We present Sparks, a multi-modal multi-agent AI model that executes the entire discovery cycle that includes hypothesis generation, experiment design and iterative refinement to develop generalizable principles and a report without human intervention. Applied to protein science, Sparks uncovered two previously unknown phenomena: (i) a length-dependent mechanical crossover whereby beta-sheet-biased peptides surpass alpha-helical ones in unfolding force beyond ~80 residues, establishing a new design principle for peptide mechanics; and (ii) a chain-length/secondary-structure stability map revealing unexpectedly robust beta-sheet-rich architectures and a "frustration zone" of high variance in mixed alpha/beta folds. These findings emerged from fully self-directed reasoning cycles that combined generative sequence design, high-accuracy structure prediction and physics-aware property models, with paired generation-and-reflection agents enforcing self-correction and reproducibility. The key result is that Sparks can independently conduct rigorous scientific inquiry and identify previously unknown scientific principles.

Figures (7)

• Figure 1: Overview of Sparks, a multi-agent AI model for automated scientific discovery. Panel a: Contemporary AI systems excel at statistical generalization within known domains, but rarely generate or validate hypotheses that extend beyond prior data, and cannot typically identify shared principles across distinct phenomena. This is because powerful models tend to memorize physics without discovering shared concepts. For scientific discovery, however, the elucidation of more general and shared foundational concepts (such as a scaling law, design principle, or crossover) is critical, in order to create significantly higher extrapolation capacity. Panel b: Sparks automates the end-to-end scientific process through four interconnected modules: 1) hypothesis generation, 2) testing, 3) refinement, and 4) documentation. The system begins with a user-defined query, which includes research goals, tools to test the hypothesis, and experimental constraints to guide the experimentation. It then formulates an innovative research idea with a testable hypothesis, followed by rigorous experimentation and refinement cycles. All findings are synthesized into a final document that captures the research objective, methodology, results, and directions for future work, in addition to a shared principle (such as in the examples presented here a scaling law or mechanistic rule). Each module is operated by specialized AI agents with clearly defined, synergistic roles.
• Figure 2: (a) Overview of the entire process from idea generation to the final document. First, the Idea Generation module formulates a high-impact research idea. Then in the testing module translates these hypotheses into executable workflows, autonomously conducting simulations or analyses to generate quantitative results. The refinement module is responsible for refining the testing strategy based on the results, adaptively revising the experimental design through an iterative feedback loop that sharpens insight and prompts reliable hypothesis testing; and writer agents consolidate the entire research lifecycle into a comprehensive document that not only presents key findings and methodologies, but also outlines future research directions—effectively serving as a blueprint for subsequent scientific inquiry. (b) Overview of the AI Agents and their role implemented in Sparks. Each module operates through a structured yet adaptive sequence of agent interactions, enabling consistency and context-aware responses across the research workflow. Each agent dynamically adapts to previous content in real time, ensuring. Inter-modular agents facilitate a generation–reflection strategy, using dynamic prompts to process evolving inputs and coordinate outputs, ensuring the system adapts fluidly to new insights throughout the research process.
• Figure 3: Overview of the Scientist_1 prompt. The prompt takes the user's query, a list of available tools, and experiment constraints as input. The agent is tasked with generating novel research ideas, guided by instructions to promote both novelty and feasibility. The agent is required to return a research idea containing several key component.
• Figure 4: Length-dependent helix–sheet mechanical crossover in short peptides. (a) User-submitted input describing the initial research query, accessible tools, experiment constraints, and total number of follow-up rounds $N_{\text{test}}$. (b) Structured research idea generated by the AI model in the Idea Generation module. (c) Multi-stage AI-driven evaluation of the hypothesis, incorporating testing and refinement. (d) Final document created by the model, provides a comprehensive overview of the research, including key results that demonstrate the length-dependent mechanical crossover—where $\beta$-sheet-biased peptides surpass $\alpha$-helix-biased peptides in unfolding force as the peptide length increases. The full document is provided in Section \ref{['sec: S1']} of the SI.
• Figure 5: Dataset overview and visualization of AI-discovered mechanical trends in short peptides. The table and plots presented here were autonomously generated by Sparks and highlight a length-dependent crossover in mechanical strength between peptide classes. In addition to producing these plots, the multi-modal model interprets the data, revealing key features such as the heterogeneity and distribution of unfolding forces. This integrated approach enhances the scientific value of the results by combining automated data analysis with interpretable visual outputs.