Skip to main content

Documentation Index

Fetch the complete documentation index at: https://mintlify.com/BDB-Genomics/AlphaGenomeR/llms.txt

Use this file to discover all available pages before exploring further.

This guide walks you through making your first multimodal genomic prediction with AlphaGenomeR. By the end, you will have queried the AlphaGenome API for a 1 MB region on chromosome 17, extracted RNA-seq and ATAC-seq predictions for lung tissue, and inspected the resulting matrix and metadata data frame. Before you begin, complete the installation steps to install the R package, the alphagenome Python dependency, and obtain your API key.

​
Run your first prediction

1

Load the library

Start a fresh R session and load AlphaGenomeR. This also loads reticulate and initializes the Python bridge on first use.
library(AlphaGenomeR)
2

Set your API key

Store your AlphaGenome API key in a variable. The recommended approach is to read it from an environment variable so it never appears in your scripts:
api_key <- Sys.getenv("ALPHAGENOME_API_KEY")
If you have not set the environment variable yet, you can assign the key directly for testing — but avoid committing this to version control:
api_key <- "YOUR_ALPHAGENOME_API_KEY"
3

Define a genomic region

AlphaGenome requires a 1 MB genomic interval in chr:start-end format using hg38 coordinates. This example uses a region on chromosome 17:
region <- "chr17:42560601-43609177"
The interval must be approximately 1 MB. The model enforces this constraint and will return an error for intervals of other sizes.
4

Query the API

Call alphagenome_query() with your API key, genomic region, tissue ontology term, and the modalities you want. This example requests RNA-seq and ATAC-seq predictions for lung tissue (UBERON:0002048):
results <- alphagenome_query(
  access_token    = api_key,
  genomic_region  = region,
  ontology_terms  = c("UBERON:0002048"),
  requested_outputs = c("RNA_SEQ", "ATAC")
)
alphagenome_query() sends a gRPC request via the AlphaGenome Python SDK and returns the full multimodal prediction as an R list. The call may take several seconds depending on the number of requested modalities.
5

Extract predictions

Use the modality-specific extractor functions to pull individual outputs from the results list:
rna_data  <- alphagenome_get_rna_seq(results)
atac_data <- alphagenome_get_atac(results)
Each extractor returns a named list with two elements:
  • $values — a numeric matrix of predictions (positions × tracks)
  • $metadata — a data frame describing each track (cell type, tissue, experiment)
6

Inspect the results

Explore the prediction matrix and track metadata:
# Dimensions of the RNA-seq prediction matrix (positions x tracks)
dim(rna_data$values)

# First few rows of the prediction matrix
head(rna_data$values)

# Track metadata: cell types, tissues, experimental details
print(rna_data$metadata)

# Plot the ATAC-seq signal for the first track
plot(atac_data$values[, 1], type = "l", main = "Predicted ATAC-seq signal")
The values matrix has one row per base position in the queried interval and one column per predicted track. The metadata data frame has one row per track, with columns identifying the tissue, cell type, and experimental conditions.

​
What’s next

Guides

Learn how to manage API keys, work with genomic regions, query tissue-specific predictions, and integrate results with Bioconductor packages.

Modalities

Explore all 11 supported genomic modalities and the extractor function for each one.

Build docs developers (and LLMs) love

Get started for freeTalk to us