Boltz-1 is a state-of-the-art open-source model that predicts biomolecular structures, including proteins, RNA, DNA, and other molecular interactions. Unlike its proprietary counterparts, it also supports modified residues, covalent ligands, and glycans, allowing for a broader range of applications. By conditioning predictions on specified interaction pockets or contacts, Boltz-1 offers a level of flexibility that makes it an indispensable tool for researchers and developers.

While AlphaFold has set a benchmark for structural prediction, Boltz-1 introduces several unique advantages. It achieves near-AlphaFold 3 accuracy while maintaining an open-access framework, broadening the scope of biomolecular modeling. Unlike AlphaFold, which focuses primarily on protein structures, Boltz-1 extends its capabilities to RNA and DNA, making it a more versatile solution for complex biological systems. Being freely available, Boltz-1 empowers a broader audience, from academic researchers to biotech startups, enabling access to cutting-edge biomolecular modeling without prohibitive costs.

The applications of Boltz-1 are diverse and impactful. It plays a crucial role in drug discovery by predicting how small molecules interact with target proteins, providing essential insights for developing new therapeutics. In synthetic biology, Boltz-1 facilitates the design of novel biomolecular structures, paving the way for innovative biotechnological solutions. Structural biology researchers rely on it to analyze complex biomolecular assemblies, while protein design specialists use it to generate stable de novo proteins.

Despite its potential, running Boltz-1 in the cloud presents several challenges. The high computational demand requires extensive GPU resources, which can be expensive and difficult to manage. Dynamic cloud pricing and fluctuating instance availability further complicate deployment, requiring expertise in cloud orchestration. Additionally, optimizing GPU selection and resource allocation is a complex task that demands specialized knowledge and extensive time, often creating a barrier to widespread adoption.

Introduction to Fovus

Fovus is an AI-powered, serverless high-performance computing (HPC) platform that delivers intelligent, scalable, and cost-efficient supercomputing power designed to enhance performance, scalability, and cost efficiency for complex HPC workloads like AlphaFold 3. Fovus uses AI to optimize HPC strategies and orchestrates cloud logistics, making cloud HPC a no-brainer and ensuring sustained time-cost optimality for digital innovation amid quickly evolving cloud infrastructure. 

Benefits of Fovus

• Free Automated Benchmarking: Evaluates how different instance choices quantitively impact the runtime and cost of your prediction computations.
• AI-Driven Strategy Optimization: Automatically identifies the best instance selection and system setup through benchmarking analysis, ensuring efficient and cost-effective computing.
• Dynamic Multi-Cloud-Region Auto-Scaling: Dynamically allocates optimal GPUs to distribute your simulations across multiple cloud regions and availability zones according to their availability dynamics. Efficiently scale up your GPU cluster and computation parallelism for biomolecular simulations.
• Intelligent Spot Instance Utilization: Intelligently utilizes spot instances according to their availability and pricing dynamics with spot-to-spot failover capability to further minimize the computation cost of biomolecular simulations.    
• Continuous Improvement: Auto-updates benchmarking data and auto-refines your HPC strategy as cloud infrastructure evolves.
• Serverless HPC Model: AI-driven automation eliminates manual setup, allowing single-command or few-click deployment. Users pay only for runtime.

One of Fovus’ key advantages is its ability to benchmark workloads on different instance choices and system configurations, providing insights into how various factors impact runtime and costs. This ensures users can make informed decisions when deploying biomolecular simulations. The platform optimizes resource allocation by dynamically selecting the most efficient instance configurations, maximizing performance, and minimizing costs. Furthermore, Fovus can distribute workloads across multiple cloud regions and availability zones, ensuring high availability and resilience.

Fovus intelligently utilizes spot instances to reduce costs further, automatically transitioning between them based on pricing and availability dynamics. This feature significantly lowers computational expenses while maintaining reliability. Additionally, the platform continuously refines its benchmarking data and HPC strategies, adapting to evolving cloud infrastructures. Its serverless architecture eliminates the need for complex manual setup, allowing users to deploy large-batch simulations with a single command while only paying for the runtime. AI autonomously optimizes performance, cost, and availability.

Optimizing Boltz-1 on Fovus

To evaluate performance, we ran Boltz-1 on Fovus using various biomolecules, each varying in complexity and size:

System	Link	Description	Atom Count	Number of Residues
System 1	7U8C	Crystal structure of Mesothelin C-terminal peptide-MORAb 15B6 FAB complex	3633	444
System 2	7BZB	Crystal structure of plant sesterterpene synthase AtTPS18	4676	554
System 3	7URD	Human PORCN in complex with LGK974 and WNT3A peptide	5534	970

We test Boltz-1 using various biomolecules on several cloud configurations. To run the workload, we used the following command: 

docker run \
    --volume=$PWD:/container_workspace \
    --workdir=/container_workspace \
    --rm \
    --ipc=host \
    --gpus all \
    fovus/boltz:latest \
    boltz predict ./input/input.fasta --use_msa_server --out_dir ./output/ --cache ./cache --sampling_steps 600 --recycling_steps 6

Boltz-1 downloads the required data and model parameters online by default, introducing runtime overhead and variance depending on network speed. Fovus hosts the required databases while also building, testing, and maintaining the Docker images for Boltz-1, enabling seamless deployment in a containerized environment. By managing all software and cloud environments, Fovus allows anyone to launch, scale, and optimize Boltz-1, which runs effortlessly out of the box and requires little to no setup. Scientists can focus on their research without the burden of dealing with software or cloud management hassles.

For each biomolecular system, the study was conducted with a specific objective for the HPC strategy optimization. The chosen objective minimized both the cost and runtime of the simulation. This strategy optimizes cost-efficiency and performance to ensure the best outcome for running simulations on the cloud.

Key performance metrics analyzed included:

• Simulation runtime
• Simulation cost

Case Study Results

Below are the performance and cost-efficiency results achieved on Fovus:

Input System	Runtime (minutes)	Cost
System 1	16	$0.10
System 2	18	$0.12
System 3	21	$0.29

The performance and cost-efficiency results achieved on Fovus demonstrate its capability to handle Boltz-1 simulations effectively. System 1 completed its run in 16 minutes, costing only $0.10. The most complex structure, System 3, finished in 21 minutes, costing $0.29.

These results highlight Fovus’ ability to optimize biomolecular prediction workflows, reducing costs while maintaining high performance. The platform efficiently distributes workloads, ensuring optimal resource usage and preventing unnecessary expenses. Additionally, as cloud infrastructure evolves, Fovus continuously refines its strategies, further improving efficiency over time.

Conclusion

Boltz-1 makes biomolecular interaction modeling more accessible, offering near-AlphaFold 3 accuracy in an open-source format. However, cloud deployment challenges such as cost, resource availability, and complexity can hinder widespread adoption. Fovus addresses these challenges by providing AI-driven optimization, intelligent auto-scaling, and cost-efficient HPC execution, ensuring a streamlined and economical solution for biomolecular simulations.

The benchmarking results demonstrate that Fovus is an ideal platform for running Boltz-1, delivering high performance at a fraction of traditional cloud costs. As cloud technologies continue to advance, Fovus will further enhance its capabilities, making biomolecular modeling even more efficient and accessible. Researchers and organizations looking to leverage Boltz-1 should consider deploying on Fovus to maximize efficiency and cost savings, accelerating scientific discovery with ease.

Explore how AI-optimized HPC can streamline your Boltz-1 workflows.