Key Concepts

Projects

Projects are the top-level organizational unit in Amazon Bio Discovery. They provide a centralized workspace for related antibody design activities.

What Projects Include

• Experiments: All computational runs and their results
• Files: Input data, structures, and generated outputs
• Collaborators: Team members with defined access levels
• Resources: Shared recipes, modules, and configurations

Project Benefits

• Organize related work in one place
• Control access and permissions
• Track resource usage and costs
• Maintain experiment history and lineage

Recipes

Recipes define computational workflows for antibody design. They specify which modules to use, how they connect, and what parameters to apply.

Recipe Types

• Hosted Recipes: Pre-built, validated workflows maintained by AWS
• Custom Recipes: User-created workflows using the recipe builder
• Shared Recipes: Custom recipes shared within organizations

Recipe Components

• Input Requirements: File types and formats needed
• Module Chain: Sequence of computational steps
• Parameters: Configurable settings for each module
• Output Specifications: Types of results generated

Common Recipe Patterns

• De Novo Design: Generate new antibodies from scratch
• Optimization: Improve existing antibody properties
• Analysis Only: Evaluate and score existing sequences
• Hybrid Workflows: Combine multiple design strategies

Experiments

Experiments are individual executions of recipes with specific input data and parameter configurations.

Experiment Lifecycle

1. Configuration: Select recipe, upload data, set parameters
2. Validation: Check inputs and estimate costs
3. Execution: Run computational workflow
4. Analysis: Review results and select candidates
5. Action: Download data or send to wet lab

Experiment Parameters

• Input Files: Target structures, seed sequences
• Design Goals: Number of candidates, optimization targets
• Constraints: Sequence regions to preserve or modify
• Scoring Criteria: Properties to evaluate and rank

Cost Considerations

Experiments consume Experiment Units (EU) based on computational complexity:

• Simple (0.5 EU): Basic analysis and small-scale design
• Standard (1.0 EU): Moderate complexity workflows
• Complex (1.5 EU): Large-scale or computationally intensive tasks

Modules

Modules are the fundamental computational building blocks that perform specific tasks in antibody design workflows.

Module Categories

• Design Modules: Generate new antibody sequences
  • De novo design algorithms
  • Directed evolution methods
  • Structure-based design tools
• Score Modules: Evaluate antibody properties
  • Binding affinity prediction
  • Developability assessment
  • Immunogenicity risk scoring
• Design and Score Modules: Combine sequence generation with property evaluation in a single workflow, enabling iterative design guided by scoring feedback
  • IntelliFold
  • Chai1
  • EvoProtGrad
  • Boltz1

Module Properties

• Inputs: Required data types and formats
• Outputs: Generated results and file types
• Parameters: Configurable settings and options
• Dependencies: Required upstream modules or data

Custom Modules / Import Modules (Beta)

Advanced users can create custom modules by:

• Containerizing algorithms using Docker
• Defining input/output schemas
• Providing test data and validation
• Publishing for team or organization use

Data Flow

Understanding how data flows through Amazon Bio Discovery helps optimize your workflows:

Input Data

• Target Structures: PDB files defining binding targets
• Seed Antibodies: FASTA sequences for optimization
• Experimental Data: Previous results for training

Processing Pipeline

1. Data validation and preprocessing
2. Module execution in defined sequence
3. Intermediate result storage and passing
4. Final result compilation and scoring

Output Data

• Candidate Sequences: Generated antibody designs
• Property Scores: Predicted characteristics
• Analysis Reports: Summaries and visualizations
• Raw Data: Detailed computational outputs

Integration Points

Wet Lab Integration

Amazon Bio Discovery connects computational design with experimental validation:

• Direct submission to partner laboratories
• Standardized assay protocols
• Result integration and analysis
• Iterative design-test cycles

External Tools

Integration with common research tools:

• Structure visualization software
• Sequence analysis platforms
• Laboratory information systems
• Data analysis and plotting tools

Cross-Region Support

Amazon Bio Discovery will automatically select the optimal region within your geography to process your inference requests. This maximizes available compute resources, model availability, and delivers the best customer experience. Your data will remain stored only in the region where the request originated, however, input prompts and output results may be processed outside that region. All data will be transmitted encrypted across Amazon's secure network.

Amazon Bio Discovery will securely route your inference requests to available compute resources within the geographic area where the request originated, as follows:

• Inference requests originating in the United States will be processed within the United States.
• If an inference request originates in an area not listed, they will be processed by default within the United States.