AlphaGenome API
Complete Developer Guide
Integrate powerful DNA sequence analysis and variant effect prediction into your applications with our Python SDK and REST API.
Quick Start
Get up and running with AlphaGenome in minutes
Install SDK
Install the Python package using pip
pip install dr7-alphagenomeSet API Key
Configure your authentication credentials
export DR7_API_KEY=your_keyRun Prediction
Make your first prediction from command line
alphagenome predict seq.fafrom dr7_alphagenome import AlphaGenome
# Initialize the client
client = AlphaGenome(api_key="your_api_key")
# Predict from DNA sequence
sequence = "ATCGATCG..." * 125000 # 1Mb sequence
results = client.predict(
sequence=sequence,
modalities=["CAGE", "ATAC", "ChIP"],
cell_types=["K562", "HepG2"]
)
# Get variant effect predictions
variant_effect = client.score_variant(
chromosome="chr1",
position=12345,
ref="A",
alt="G"
)
print(f"Variant effect score: {variant_effect.score}")
print(f"Affected tracks: {variant_effect.top_affected_tracks}")Core API Capabilities
Powerful endpoints for comprehensive genomics analysis
Sequence Prediction
POST /v1/alphagenome/predict
Predict functional effects from raw DNA sequences. Returns predictions across 5,930 tracks covering gene expression, chromatin accessibility, and more.
Parameters:
- •sequence: DNA sequence (up to 1Mb)
- •modalities: List of output types (CAGE, ATAC, etc.)
- •cell_types: Target cell types/tissues
- •output_format: json, bigwig, or bedgraph
Variant Scoring
POST /v1/alphagenome/variant
Score the functional impact of genetic variants. Compare reference and alternate allele predictions to quantify variant effects.
Parameters:
- •chromosome: Chromosome name (chr1-chr22, chrX, chrY)
- •position: Genomic position (hg38)
- •ref: Reference allele
- •alt: Alternate allele
- •window_size: Context window (default 100kb)
Batch Processing
POST /v1/alphagenome/batch
Process multiple sequences or variants in a single request. Ideal for VCF files or genome-wide studies.
Parameters:
- •variants: List of variants or VCF file
- •parallel: Enable parallel processing
- •callback_url: Webhook for completion notification
- •output_format: Output file format
Region Query
GET /v1/alphagenome/region
Query predictions for a specific genomic region. Useful for visualizing predictions in a genomic browser.
Parameters:
- •chromosome: Target chromosome
- •start: Start position
- •end: End position (max 1Mb range)
- •tracks: Specific tracks to return
Code Examples
Learn by example with these common use cases
Basic Sequence Prediction
from dr7_alphagenome import AlphaGenome
client = AlphaGenome()
# Load sequence from FASTA file
results = client.predict_from_fasta(
"input.fa",
modalities=["CAGE", "ATAC", "DNase", "H3K27ac"],
output_format="bigwig"
)
# Results are saved as BigWig files for visualization
for track in results.tracks:
print(f"Saved: {track.output_path}")Variant Effect Scoring
from dr7_alphagenome import AlphaGenome
client = AlphaGenome()
# Score a single variant
effect = client.score_variant(
chromosome="chr17",
position=7674220, # Near TP53 gene
ref="G",
alt="A",
window_size=100000 # Context around variant
)
print(f"Overall effect score: {effect.score:.4f}")
print(f"Gene expression change: {effect.expression_delta:.4f}")
print("Top affected cell types:")
for ct in effect.top_cell_types[:5]:
print(f" {ct.name}: {ct.score:.4f}")Batch VCF Processing
from dr7_alphagenome import AlphaGenome
client = AlphaGenome()
# Score variants from VCF file
results = client.score_vcf(
"variants.vcf",
reference="hg38",
output_file="scored_variants.tsv",
include_tracks=["CAGE", "enhancer_activity"],
parallel=True
)
print(f"Scored {results.total_variants} variants")
print(f"High-impact variants: {results.high_impact_count}")
# Filter for regulatory variants
regulatory = results.filter(
min_score=0.5,
affected_elements=["enhancer", "promoter"]
)
regulatory.to_dataframe().to_csv("regulatory_variants.csv")REST API with curl
# REST API Example with curl
# Predict from sequence
curl -X POST https://api.dr7.ai/v1/alphagenome/predict \
-H "Authorization: Bearer $DR7_API_KEY" \
-H "Content-Type: application/json" \
-d '{
"sequence": "ATCGATCG...",
"modalities": ["CAGE", "ATAC"],
"cell_types": ["K562"]
}'
# Score variant
curl -X POST https://api.dr7.ai/v1/alphagenome/variant \
-H "Authorization: Bearer $DR7_API_KEY" \
-H "Content-Type: application/json" \
-d '{
"chromosome": "chr1",
"position": 12345,
"ref": "A",
"alt": "G"
}'Need Help with the API?
Ask our AI assistant for help with code examples, troubleshooting, or integration questions
AlphaGenome API Assistant API Technical Support
Experience Genomics AI
Best Practices
Optimize your AlphaGenome API usage
Sequence Length Optimization
Balance context length with processing time for optimal results.
- ✓Use 100kb-500kb windows for most applications
- ✓Full 1Mb context for long-range regulatory analysis
- ✓Smaller windows (50kb) for high-throughput variant scoring
Cell Type Selection
Choose relevant cell types to focus your analysis.
- ✓Select cell types matching your biological question
- ✓Use tissue-matched cell types for disease studies
- ✓Query available cell types with /v1/alphagenome/cell-types
Batch Processing Strategy
Efficiently process large datasets with batch endpoints.
- ✓Group variants by chromosome for efficient processing
- ✓Use async callbacks for large jobs (>1000 variants)
- ✓Consider output format based on downstream analysis
Error Handling
Build robust applications with proper error handling.
- ✓Implement exponential backoff for rate limits
- ✓Validate sequences before submission
- ✓Check variant positions against reference genome
Ready to Scale Your Genomics Analysis?
Get API access with higher rate limits, priority support, and advanced features.
API FAQ
Common technical questions about the AlphaGenome API
What are the rate limits?
Free tier: 100 requests/day. Standard: 10,000 requests/day. Pro: 100,000 requests/day. Enterprise: Custom limits available. Rate limit headers are included in all API responses.
What input formats are supported?
The API accepts raw DNA sequences (A, T, C, G characters), FASTA format, genomic coordinates (hg38), and VCF files for batch variant scoring. Sequences can be up to 1Mb in length.
How do I handle long sequences?
For sequences longer than 1Mb, split them into overlapping windows and combine predictions. The SDK provides a sliding_window() helper function that handles this automatically.
What output formats are available?
JSON (default) for programmatic access, BigWig for genome browser visualization, BedGraph for text-based analysis, and TSV for tabular outputs. Batch jobs can also output compressed files.
How do I authenticate?
Use Bearer token authentication with your API key. Set the DR7_API_KEY environment variable or pass it directly to the client constructor. API keys can be managed in your Dr7.ai dashboard.
Is there a Python SDK available?
Yes! Install with 'pip install dr7-alphagenome'. The SDK provides a high-level interface for all API endpoints, automatic retries, streaming support, and helper functions for common workflows.