AWS Batch Integration

Scale your drug discovery workloads across thousands of compute instances with seamless AWS Batch integration.

Overview

Drug discovery computations—virtual screening, molecular dynamics, digital patient simulations—often require massive parallel compute resources. Omic's AWS Batch integration enables you to scale these workloads elastically without managing infrastructure.

Enterprise customers can connect their own AWS accounts for compute, maintaining data sovereignty while leveraging Omic's orchestration layer. Pay only for the compute you use.

Architecture

Omic Platform

Job orchestration & monitoring

↓

AWS Batch

Job scheduling & compute management

↓

EC2 Spot

EC2 GPU

Graviton

Omic handles job scheduling, retry logic, spot instance interruption handling, and result aggregation. You define the compute requirements; we manage the complexity.

Use Cases

Ultra-Large Virtual Screening

Screen billions of compounds against multiple targets. AWS Batch auto-scales to thousands of instances for massive parallelism.

Molecular Dynamics

Run long MD simulations with GPU acceleration. Automatic checkpointing handles spot instance interruptions.

Digital Patient Cohorts

Simulate millions of virtual patients in parallel for population-scale drug response prediction.

Generative Chemistry

Train and run generative models on GPU clusters for de novo compound design and optimization.

Usage Example

from omic import Omic
from omic.compute import AWSBatchConfig

client = Omic(api_key="your_api_key")

# Configure AWS Batch compute environment
compute_config = AWSBatchConfig(
    # Use Omic's managed environment (default)
    # Or connect your own AWS account:
    # aws_account_id="123456789012",
    # aws_region="us-east-1",

    instance_types=["c6i.8xlarge", "c6i.16xlarge", "m6i.8xlarge"],
    use_spot=True,
    spot_bid_percent=70,  # Bid 70% of on-demand price
    max_vcpus=10000,
    gpu_instance_types=["p4d.24xlarge", "g5.12xlarge"],  # For ML workloads
)

# Run large-scale virtual screening
screening_job = client.virtual_screen(
    target="P00533",  # EGFR
    library="enamine_real_10B",  # 10 billion compounds
    compute=compute_config,
    parallelism=5000,  # Run on 5000 instances
    chunk_size=100000  # 100K compounds per job
)

# Monitor progress
screening_job.subscribe_progress(
    callback=lambda p: print(f"Progress: {p.percent:.1f}% "
                              f"({p.compounds_screened:,} compounds, "
                              f"{p.active_instances} instances)")
)

# Wait for completion
results = screening_job.wait()

print(f"Screening complete!")
print(f"  Total compounds: {results.total_compounds:,}")
print(f"  Compute time: {results.total_compute_hours:.1f} hours")
print(f"  Cost: {'$'}{results.total_cost:.2f}")
print(f"  Hits found: {results.num_hits:,}")

Bring Your Own Cloud (BYOC)

Enterprise customers can connect their own AWS accounts to maintain full data sovereignty while using Omic's orchestration:

Data never leaves your AWS account
Use your existing AWS credits and reserved instances
Integrate with your VPC, security groups, and IAM policies
Audit logs in your CloudTrail
Compliance with your organization's cloud policies

# Connect your AWS account
from omic.compute import connect_aws_account

connection = connect_aws_account(
    aws_account_id="123456789012",
    external_id="omic-integration-abc123",
    role_arn="arn:aws:iam::123456789012:role/OmicBatchRole",
    s3_bucket="my-company-omic-data",
    vpc_id="vpc-0abc123def456",
    subnets=["subnet-aaa", "subnet-bbb"],
    security_groups=["sg-omic-compute"]
)

# Verify connection
connection.verify()
print(f"Connected to AWS account {connection.account_id}")

Pricing

Compute Type	Managed (Omic Cloud)	BYOC
CPU (per vCPU-hour)	$0.05	Your AWS cost + $0.01
GPU (per GPU-hour)	$2.50	Your AWS cost + $0.25
Spot instances	60-80% discount	Your spot pricing

Scale your drug discovery compute

Run billion-compound screens and population-scale simulations with elastic cloud compute.