abYcloud: Advancing Antibody Therapeutics Research with a user-friendly platform powered by AI

abYcloud is revolutionising antibody drug development through its AI-powered service, Axiom, by drastically reducing the time required to develop new antibody drugs from years to mere weeks.

Founded by Shibo Wu and Chu'nan Liu from UCL's School of Pharmacy and Research Department of Structural Molecular Biology, abYcloud’s SaaS platform leverages advanced data analytics and machine learning to streamline research processes, with the potential to have a transformative impact on drug discovery and optimisation. Chu'nan Liu, a recipient of the CDI AWS Doctoral Scholarship in Digital Innovation, was already immersed in the CDI community when he and his co-founder embarked on the journey to translate their research into a viable commercial venture. Their progression from early-stage support to acceptance into the fifth cohort of the UCL CDI Impact Accelerator epitomises a success story illustrating CDI's pivotal role in nurturing digital innovation and catalysing the acceleration of research into a digital solution which has the potential to significantly impact public health.

Challenges in Antibody Therapeutics

Developing antibody therapeutics is critical due to antibodies' pivotal role in the immune system's ability to identify and neutralise pathogens effectively. Through engineering, antibodies can be customised to target specific diseases with greater precision than conventional treatments, promising significant advancements in therapy and improving patient outcomes. However, this research is fraught with challenges, including high costs and lengthy timelines. According to Deloitte (2022), developing new therapies for cancer treatment can cost approximately £2.2 billion and take 5 to 15 years. Traditional methods are slow, expensive, and reliant on a trial-and-error approach.

Customising antibodies for each patient adds complexity to optimising antibody treatments within personalised medicine approaches. Even within the same type of cancer therapy, adjustments are necessary based on patients' genetic profiles. Achieving specificity and efficacy can be time-intensive, often taking over six months using conventional techniques. Consequently, there is an urgent need for more efficient and precise methods to accelerate the antibody development workflow, enhance research efficiency, and enable scientists to concentrate on the most impactful aspects of their work with heightened accuracy.

The Solution

Building on their PhD research, the co-founders of abYcloud have transformed their expertise into a pioneering commercial venture with the development of the Axiom platform. This innovative solution leverages advanced data analytics and AI technologies to revolutionise the development process of

antibody drugs. Axiom integrates high-quality antibody databases with novel AI methodologies, providing researchers and pharmaceutical companies worldwide with seamless access to powerful tools that promise to expedite the creation of life-saving therapies.

At its core, Axiom offers three essential service pillars, each poised to drive significant social impact:

• Data Cleaning Services: Ensures researchers can annotate and prepare their data accurately for analysis. By standardising and refining data inputs, this service accelerates the preliminary stages of drug development, allowing scientists to focus on high-value tasks. This enhancement not only saves time but also maximises the use of valuable research data, paving the way for faster drug discoveries.
• Artificial Intelligence: Delivers predictive insights and structural recommendations, guiding the selection of optimal drug candidates and the enhancement of existing therapies. These AI-driven approaches can identify promising candidates far more efficiently than traditional methods, dramatically reducing the time and cost associated with drug discovery. This increased efficiency holds the potential to bring critical treatments to market much sooner, directly benefiting patients awaiting new therapies.
• Data Visualisation Services: Provides intuitive data visualisation, offering a 3D map-like view of antibody-antigen interactions that clarifies molecular mechanisms and supports the creation of more effective antibodies. By making complex data accessible and comprehensible, these visualisations empower researchers to make informed decisions swiftly. This clarity accelerates therapeutic antibody refinement, enhancing their effectiveness and specificity in targeting diseases.

Technical Advancements with the CDI

Scaling Through Serverless Architecture

During their participation in the CDI Impact Accelerator (IA), abYcloud underwent a strategic evolution and transformed from a locally run Minimum Viable Product (MVP) to a robust, scalable web application by switching to a serverless architecture on Amazon Web Services (AWS). They moved their AI models and bioinformatics pipelines from local containers to Amazon ECS and AWS Batch, and redesigned their workflow using AWS Lambda, Step Functions, and DynamoDB. This change improved scalability, reliability, and cost-effectiveness. The new infrastructure not only bolsters current operations but also positions abYcloud for future growth, ensuring researchers worldwide instant access to cutting-edge AI models and resources.

Enhanced Functionality in Data Handling

Simultaneously, abYcloud focused on implementing core functionalities critical to their platform's success. They integrated sequence and structure cleaning services to ensure precise data annotation for analysis. This enhancement facilitated seamless job submission processes, empowering researchers to effectively utilise AI-driven insights in antibody drug development.

Development of Web-Based User Interface (UI)

Driven by a commitment to user accessibility and security, abYcloud developed a sophisticated web-based UI using AWS Amplify and AWS Cognito. This effort significantly enhanced the platform's user management capabilities and integrated advanced visualisation tools. By migrating structure visualisation to the web UI and introducing features like therapeutics and drug target interface region highlighting, abYcloud enriched user interaction and analytical depth.

Adoption of Infrastructure as Code

Early in the IA programme, abYcloud participated in a CDI workshop focused on "DevOps: Infrastructure as Code." Armed with this knowledge, they adopted AWS CDK (Cloud Development Kit) to streamline platform infrastructure provisioning. This approach afforded them flexibility in deploying and managing resources, optimising development cycles, and operational efficiency.

Impact and Future Directions

Following the Impact Accelerator, abYcloud conducted extensive user testing and received overwhelmingly positive feedback that validated the platform's functionality and user experience. Their technical advancements significantly accelerated antibody therapeutic candidate identification, and in one use case, they proved to reduce timelines from 2-3 weeks to just 2 days. By collaborating with lab researchers and leveraging AWS support, abYcloud aims to further streamline research using AWS Bedrock’s knowledge base and Retrieval Augmented Generation (RAG)-based approaches.

The Axiom platform's impact extends beyond laboratories, with the potential to transform drug discovery by shortening processes from months to mere weeks. This increased speed allows for quick responses to health threats, potentially saving lives with timely treatments. Additionally in educational settings, Axiom's interactive visualizations are essential for training future biomedical researchers, enhancing their understanding of complex drug optimization mechanisms and fostering a skilled workforce for healthcare innovation.

abYcloud's platform integrates cutting-edge technology with scientific research to provide tangible societal benefits, advancing antibody therapeutics and promoting global health resilience.

Co-Founder, Chu'nan Liu, summarises the team's experience with the CDI:

"The CDI Impact Accelerator program has provided us with an excellent foundation, enabling the development of our Minimum Viable Product (MVP)—a Software as a Service (SaaS) platform that equips lab researchers with advanced AI models. We've benefited immensely from the guidance of experts at the Advanced Research Computing Centre at UCL and Solution Architects from AWS. Our MVP harnesses AWS's capabilities effectively, utilizing serverless architecture for flexible job handling, AWS Amplify for seamless website hosting and integration with other AWS services, and AWS Step Functions along with AWS Batch to manage complex job logic efficiently. We've also gained valuable business insights, such as Amazon's Leadership Principles, which have profoundly influenced our approach to business and product development, emphasizing a user-centric strategy."