Principle of Integrated Human Practice

Find out more about what we've done to advance our project!

How do we understand the local challenges

---

Our investigation into antimicrobial peptide (AMP) research revealed several fundamental limitations that impede data integration and downstream analysis.

AMP-related information remains highly fragmented across heterogeneous repositories, each employing distinct annotation schemes, terminologies, and experimental standards. Such inconsistency compromises data interoperability and undermines cross-database comparison.

Moreover, essential biological descriptors, including activity, stability, and toxicity, are frequently incomplete or ambiguously defined, resulting in unreliable modeling outcomes and limited reproducibility.

The retrieval of compound biological information is equally constrained by the absence of semantic and multi-parameter query mechanisms, restricting systematic knowledge extraction.In addition, experimental validation of uncharacterized peptide sequences entails considerable cost and time, further constraining progress from in silico prediction to empirical verification.

Collectively, these limitations delineate a pronounced data bottleneck that hinders both algorithmic innovation and translational advancement, despite the escalating clinical demand for effective antimicrobial alternatives.

IBI EXPO: A Journey into the Synthetic Biology Industry

Our team participated in the 2025 International Bio-Industry Expo (IBI EXPO). As a premier global biotechnology event, this year's IBI EXPO, themed "Synergizing Cutting-Edge Technologies, Fostering Industrial Convergence and Innovation," featured four core exhibition areas: Biopharmaceutical R&D, Intelligent Medical Devices, Synthetic Biology Technologies, and Bioinformatics Tools.

It attracted over 800 participants from more than 30 countries and regions — including showcases of the latest R&D achievements from multinational pharmaceutical companies like Pfizer and Roche, technology sharing from research institutions such as Germany's Max Planck Institute and the National University of Singapore, and case studies on industrial application from domestic enterprises like BGI and Suzhou Mediphage Bio.

During the expo, we engaged in an in-depth discussion with Dr. Chen, co-founder of Suzhou Mediphage Bio, one of the few enterprises advancing the commercialization of antimicrobial peptides. His professional perspective underscored the current stagnation of AMP translation, noting that only a handful of candidates have progressed beyond laboratory evaluation due to data fragmentation, inconsistent characterization, and limited predictive tools. These observations directly resonated with our preliminary analysis of the field and reinforced our commitment to developing SPADE—a systematic, data-driven platform designed to unify AMP resources and provide computational support for efficient peptide discovery and evaluation.

Post-Discussion with Professor Xu Dechang: New Reflections on the Feasibility of Our iGEM AMP Project

To address our uncertainties in database construction, our team consulted Professor Xu Dechang on the feasibility of our antimicrobial peptide (AMP) project, focusing on two key technical directions: building a RAG-enhanced retrieval model and using large language models for AMP prediction.

Regarding the RAG model, we presented our plan to integrate authoritative databases like UniProt and process documents using fixed and content-based chunking. Professor Xu emphasized the importance of vector database selection and annotating critical attributes, particularly highlighting biologically relevant information such as "antimicrobial activity" and "toxicity."

When discussing AMP prediction, we outlined our approach to fine-tune a GPT-based model on AMP datasets. He recommended experimenting with multiple models for comparative analysis and selecting the best-performing one.

This conversation provided clear, actionable pathways for our originally theoretical technical plan, reinforcing our confidence in the project's feasibility.

Defining an Effective Solution

---

Addressing the challenge of limited and complex antimicrobial peptide (AMP) information, SPADE serves as an optimal open-source AMP database. Its open-access nature breaks down data barriers, allowing researchers to freely obtain and share core data—such as sequences, activity, and mechanisms—without the constraints of commercial licenses or costs. This significantly reduces initial R&D expenses and supports customizable secondary development for enhanced flexibility, thereby boosting research efficiency.

Concurrently, its transparent data resources and user-friendly access enable non-specialists, including students and science enthusiasts, to easily explore standardized AMP information. This not only facilitates fundamental understanding but also deepens awareness of the value and application of biological data, effectively fulfilling the database's role in both research and public education.

Discussion with Professor Song Hao: Laboratory Application of Antimicrobial Peptides

Professor Song Hao, with his profound expertise in synthetic biology, provided valuable insights during our discussion on antimicrobial peptide (AMP) applications. He highlighted that synthetic biology approaches could significantly advance laboratory research on AMPs, thereby strengthening the theoretical foundation for their use in medicine—such as developing novel antibacterial agents to combat drug-resistant bacteria—and in agriculture, where AMPs could replace certain antibiotics in managing plant and animal diseases.

Professor Song also emphasized the importance of addressing AMP stability and specificity, as well as prioritizing user experience in database design. He expressed interest in utilizing our product upon project completion.

This exchange provided multi-dimensional perspectives on AMP application and real-world user needs, guiding our dry-lab team to focus on key predictive parameters—including MIC, half-life, and hemolytic activity—and enhancing the biological relevance of our project design.

Professor Wang Ruiyao's Insight: Key Trade-offs in Antimicrobial Peptide Database Construction

In discussing the construction of an antimicrobial peptide database, Professor Wang Ruiyao emphasized physicochemical properties as the core foundation. He stressed the necessity of including key parameters like isoelectric point, hydrophobicity, peptide length, and solubility, noting that their absence would severely hinder candidate screening and experimental design, thus diminishing the database's utility.

While acknowledging the value of molecular dynamics simulations for visualizing peptide-target interactions, he pragmatically advised against their immediate inclusion due to limited structural data and a lack of standardized protocols, which could compromise database stability. His guidance was to first solidify the core dataset and postpone advanced features for a future upgrade, thereby clarifying priorities and teaching us to balance functional aspirations with practical constraints.

Bio-China Expo: Connecting with Cutting-Edge Industry Resources

To broaden industry connections and secure project support, our team actively participated in the 2025 Bio-China Expo. Guided by the principle of "two-way communication and targeted engagement," we adopted a dual approach: On one hand, we delivered systematic presentations to exhibitors in biopharmaceuticals and synthetic biology, outlining our project vision, core technologies, and implementation needs to explore collaboration opportunities based on technical compatibility and resource synergy.

On the other hand, we conducted comprehensive field research through exhibition visits, forum participation, and in-depth discussions with R&D leaders to gather insights on technological advances, industry challenges, and market trends — providing critical references for project optimization. These efforts enabled us to gain a realistic understanding of industry demands.

Discussion with Dr. Yu Qiang: Key Recommendations for Building an Antimicrobial Peptide Database

During our exchange with Dr. Yu Qiang of Shengshi Taike, he provided clear, practical guidance that steered our project in a focused direction. While he did not delve deeply into the technical specifics we prepared, he emphasized two critical points:

He first acknowledged the value of a database in optimizing resources and improving research efficiency, but then highlighted the core challenge: "With limited studies and sparse foundational data on antimicrobial peptides (AMPs), the accuracy of any conclusions will be constrained."

Dr. Yu further stressed that beyond key drugability indicators—such as antimicrobial activity, toxicity, and stability—other dimensions like physicochemical properties, source organisms, antimicrobial spectrum, and mechanisms of action are equally important. He specifically advised: "Only by refining these dimensions into specific, quantifiable metrics can the database truly support user decision-making."

His straightforward insights shifted our focus from purely technical framework design to addressing fundamental issues of data accumulation and metric quantification. As a result, following the refinement of our technical architecture, we pivoted from initially planning to integrate a GPT-based model to developing our own Transformer model, tailored for predicting specific AMP properties.

How do we implement

Our solution builds upon the foundational framework of the SPADE database, which provides large-scale, systematically curated AMP resources. To extend its analytical capacity, we integrated two major components: the AMP-Oriented Multi-task Model (AOMM) and a RAG-enhanced retrieval module.

The AOMM model was specifically developed to address the bioinformatics challenge presented by the AMP-Oriented Multi-Property Prediction Task (AMPPT) within the AMPOS dataset. It achieves comprehensive evaluation of antimicrobial peptides through a unified neural network framework encompassing five subtasks: sequence mask prediction, AMP classification, half-life regression, microorganism-specific MIC regression, and hemolytic activity regression. Demonstrating state-of-the-art performance across all subtasks, AOMM enables robust, multi-dimensional property inference for both natural and modified peptides.

Complementarily, the RAG module enhances the precision and efficiency of complex information retrieval by leveraging SPADE's structured data on physicochemical and mechanistic properties. Together, these components form an integrated "retrieval + prediction" system, advancing SPADE from a static data repository to an intelligent platform for AMP discovery and evaluation.

Follow-up Discussion with Professor Xu Dechang: RAG Model Validation and New Project Directions

Equipped with our team's fully developed RAG model and comprehensive RAG validation data, we held another meeting with Professor Xu Dechang. During the presentation, we demonstrated the model's accuracy in retrieving antimicrobial peptide data—comparative evaluations across multiple vector databases and RAG validation experiments indicated an 82% response accuracy under masked training conditions, confirming its ability to effectively support fundamental data retrieval tasks.

After reviewing the validation report, Professor Xu recognized the model's practical progress and offered valuable suggestions for further enhancement. He recommended extending the model's scope from data retrieval toward the application-oriented transformation of antimicrobial peptides. Moreover, he emphasized the essential distinction between linear and cyclic peptides, noting their considerable differences in stability, antimicrobial spectra, and mechanisms of action. To facilitate more targeted screening, he proposed adding a "peptide chain structural type" annotation to the existing physicochemical property indicators.

Given the time constraints of the iGEM competition, full implementation of this recommendation may not be achievable within the current cycle. Nevertheless, the concept of developing an "Antimicrobial Peptide Evolver" represents a promising direction for future advancement of the project.

Expanding Our Horizons at CMC PharmExpo: Connecting Database Development with Industry Realities

As an iGEM team dedicated to developing an antimicrobial peptide database, we recently participated in the CMC PharmExpo, where we gained valuable industry insights by attending enterprise seminars and engaging in meaningful dialogues with over three peptide-focused companies. This experience provided us with first-hand exposure to the current landscape of peptide drug development, including key challenges such as production scalability, modification techniques, and clinical translation pathways.

Through strategic conversations with industry representatives, we identified specific gaps where our database could serve practical needs—such as providing accessible activity data and stability parameters to assist in early-stage peptide screening. These interactions also underscored the importance of ensuring our database remains user-centered, offering clear, actionable information that can accelerate both academic and industrial research.

This experience has reinforced our commitment to creating not just a repository of data, but a functional platform that bridges wet-lab research and real-world application. Moving forward, we are more motivated than ever to refine our database with industry-informed features, ensuring its practical relevance and supporting the broader peptide community in developing solutions to antimicrobial resistance.

Exchanging and Broadening Horizons: Offline Collaboration with Fellow Teams

On September 21, our team joined forces with two dedicated synthetic biology teams—East China University of Science and Technology's Wukong Team and Nanjing Tech University's NJTECH-CHINA-A—for an efficient and pragmatic dialogue centered on "sharing experiences and solving challenges together." This collaboration injected fresh momentum into all three teams' iGEM project development and competition preparation.

From the outset, the teams focused on core issues spanning the entire iGEM project lifecycle—from topic selection and experimental execution to results optimization. Each team presented their work, openly discussed obstacles, raised questions, and jointly explored practical solutions.

This cross-university exchange not only helped clarify each project's optimization pathway but also fostered a consensus on sustained collaboration. Moving forward, the teams will establish a regular communication mechanism to share resources and tackle challenges aligned with key iGEM milestones.

By embracing this "complementary and collaborative" model, we have moved beyond merely completing our project—we have built meaningful connections with the wider iGEM community. This experience strengthened our sense of belonging and offered a profound sense of achievement: through educating and being educated, we witnessed how teamwork can lead to tangible, positive impacts.

Online Meetings: A Bridge for Humanistic Care

On September 22, our team jointly held an online exchange session with multiple teams, including one from Tongji University, centered on the theme of "Humanistic Care." This meeting moved beyond simple technical project sharing, instead focusing on conveying human warmth in scientific research and practice, aiming to explore the deep integration of "project value" and "societal needs."

During the exchange, various teams shared multidimensional perspectives on "implementing humanistic care in projects":

• The Tongji University team, using their crayfish project as an example, explained how they grounded their work in reality by understanding astronauts' actual needs.
• The Nanjing Tech University team shared their development of dressings for pets, highlighting the problem of prohibitively high costs in pet healthcare.
• Our team identified the escalating challenge of antimicrobial resistance and its profound implications for global health, emphasizing the absence of efficient peptide screening and evaluation frameworks during the early stages of antimicrobial peptide (AMP) research and development. In response, we proposed our database-driven initiative as a systematic solution designed to enhance AMP discovery efficiency and support the rational design of novel therapeutic candidates.

The session consistently emphasized the core principle of "not doing projects just for the sake of projects." All teams agreed that scientific research and project advancement should prioritize solving real-world problems and serving specific groups, rather than merely pursuing technical metrics or formal outcomes. For instance, when discussing project optimization, the consensus was to prioritize user-friendliness and needs-matching over blindly adopting complex technologies.

How do we evaluate

Following the refinement of the project plan under instructor Han's guidance, the project was formally deployed to address real industry pain points. It received positive feedback from relevant sectors and was well received at events such as the Jiangsu-Zhejiang-Shanghai Exchange Forum and the CCIC activity. Subsequently, we promoted and introduced the project within universities, while also engaging in mutual testing and feedback exchanges with peers in the same field.

Organizing the Jiangsu-Zhejiang-Shanghai iGEM Exchange: Platform Construction and Realizing the Value of Communication in Synthetic Biology

During the internal project presentation session, each team delivered a structured academic report summarizing their progress, encompassing experimental design principles, optimization of technical workflows, and strategies for overcoming key bottlenecks. This format facilitated in-depth exchange of research concepts and practical experience within the community.

In the external exhibition area, an "academic marketplace" model was adopted, encouraging participants to engage in focused, poster-based discussions on specific technical topics, including module construction logic and sample processing methodologies. This interactive approach effectively transformed online collaborations into tangible offline academic partnerships, thereby strengthening direct communication and long-term cooperation among teams.

In terms of exchange outcomes, the academic value and organizational impact of the event exceeded expectations. On one hand, through cross-team technical discussions, some participants gained solutions to experimental challenges, clarified their project pathways, and enhanced their motivation and confidence in participating in the iGEM competition. On the other hand, thematic discussions on the practical applications of synthetic biology shifted participants' understanding of the discipline from theoretical concepts to real-world research. This not only expanded the academic influence of the iGEM competition in the Jiangsu-Zhejiang-Shanghai region but also deepened participants' comprehension of the research paradigms in synthetic biology.

Our most significant contribution lies in transforming a one-off event into a prototype of a sustainable academic community. By fostering a collaborative atmosphere grounded in shared interests and goals, we have effectively facilitated a shift in the mindset of iGEM participants—from "competitors" to "collaborators."

In the concluding phase of the event, the active feedback from attendees affirmed the necessity of such a platform. As organizers, we deeply recognize that this exchange was not merely an isolated activity, but an important vehicle for building a regional iGEM academic community. The connections formed through common research interests will provide ongoing academic support and enable resource complementarity among teams in their future projects.

Through this event, we promoted our own project while also engaging in mutual discussions about the value and needs of each other's work. This allowed us not only to gain recognition but also to experience the meaningful impact of human initiative—ensuring that our projects genuinely serve society and bring positive influence to our environment and communities. Organizing and participating in this activity enabled us to receive feedback on our own work, while also creating a chain of support that helped other iGEM community members exchange ideas and grow together.

Excelling in the CCiC Software Track: Our Team's Antimicrobial Peptide Database Project Earns Synbiopunk Third Prize in China

During the CCiC conference, our team delivered an in-depth presentation on the development of our Antimicrobial Peptide Database (SPADE). The presentation systematically introduced the project's integrated framework, encompassing the RAG-enhanced retrieval model, the AMP-Oriented Multi-task Model (AOMM) for multi-property prediction, and the data standardization pipeline that unifies and quantifies key metrics such as physicochemical properties and antimicrobial spectra. Together, these components demonstrated the platform's capacity for intelligent retrieval and comprehensive evaluation of antimicrobial peptides. This presentation effectively conveyed the project's technical robustness, scientific rigor, and its potential to serve as a foundational infrastructure for AI-driven peptide research.

In the Software track pitch session, our core concept of "accelerating antimicrobial peptide R&D through a data-driven approach" was highly endorsed by the judges and attendees. This recognition was further supported by the project's demonstrated strengths in data integration (covering over 39,000 antimicrobial peptide entries) and functional completeness (featuring multi-dimensional search and preliminary activity prediction).

Ultimately, the project was awarded the Third Prize in the public voting segment at the Synbiopunk iGEMer Exchange Conference in China. This achievement validates the project's practical value and the soundness of its scientific objectives within the field of synthetic biology applications.

Picture Book Project: Making Science Accessible to Children

Amid the growing global issue of antibiotic resistance and the general public's limited understanding of antimicrobial peptides, the XJTLU-Software team from Xi'an Jiaotong-Liverpool University and the NJTECH-CHINA-A team from Nanjing University of Technology leveraged the "HP Activity" as an opportunity to collaboratively create a series of popular science picture books on antimicrobial peptides, combining scientific value with humanistic concern. During the activity, the teams from both universities efficiently collaborated by leveraging their disciplinary strengths: the Nanjing University of Technology team, drawing on its extensive research experience, precisely outlined the core knowledge on "the definition of antimicrobial peptides, their mechanisms of action, differences from antibiotics, and application scenarios," thereby ensuring the scientific accuracy of the content; the Xi'an Jiaotong-Liverpool University team focused on the 7-15 age group, designing anthropomorphic characters such as the "Blue Little Guardian (antimicrobial peptide)" and the "Gray Mischief Bug (bacteria)," using bright illustrations and narrative storytelling to connect knowledge points and make complex concepts more easily understandable.

The resulting popular science picture books not only break down barriers to the dissemination of antimicrobial peptide knowledge but also provide a high-quality medium to enhance scientific literacy among young people within the framework of the "HP Activity." Furthermore, the project sets a model for inter-university scientific collaboration. Moving forward, the teams plan to further broaden the reach of these educational resources through multi-scenario promotion and digital version development. Both universities share a common goal: they are committed not only to completing the project but also to fostering a sense of humanistic care, aiming to expand the influence of antimicrobial peptides through picture books, reduce the hazards of antibiotic resistance, and attract more people to understand and engage in scientific endeavors.

Feedback: Exchanging Insights with a Team in the Same Antimicrobial Peptide Field

On September 30, our team held a productive online meeting with Beijing Institute of Technology to discuss research project advancement and the application of technical platforms. To ensure efficient communication, both teams shared their official websites in advance, allowing preliminary understanding through project introductions and technical documentation, and preparing for subsequent platform testing discussions—laying a solid foundation for in-depth exchange.

At the start of the meeting, both sides shared updates on their core projects: our team elaborated on the current technical framework, key research directions, and latest progress in experimental data collection and成果转化；the BIT team also outlined their project developments in synthetic biology, highlighting technical challenges and phased breakthroughs. Subsequently, both teams exchanged feedback on tool platform usage: regarding the SPADE antimicrobial peptide database we use, they provided optimization suggestions such as improving data retrieval convenience and information update timeliness. The BIT team shared their experience using our functional modules and introduced the operational workflow and advantages of the FABRIC platform, offering specific references for further platform refinement and collaboration.

Engaging with peers in the same field not only deepens our own domain expertise but also facilitates valuable learning from others' achievements.

Participating in Alibaba Cloud's Yunqi Conference: Connecting Our Model with China's AI Frontier

As a representative of our team, I had the privilege of attending the Alibaba Cloud Yunqi Conference, a premier gathering for innovation in China. A key highlight was the opportunity to engage with leading AI entrepreneurs and present the core model behind our project.

We initiated in-depth discussions with founders and technical leads from several cutting-edge AI companies. The focus was on our proprietary model—specifically, its architecture, training methodology, and its application in predicting key peptide properties. These exchanges provided invaluable, real-world perspectives on the scalability and practical deployment of computational biology tools. The feedback was immensely constructive, challenging us to consider optimization pathways and potential integration scenarios we had not previously explored.

This experience was pivotal. It allowed us to validate our technical approach against industry benchmarks and solidified our conviction that our model addresses a tangible gap at the intersection of AI and biotechnology. The connections forged at Yunqi will undoubtedly serve as a foundation for future collaboration, driving our project from a promising prototype toward a tool with real-world impact.