Alphagenome Tools

Configuration File: alphagenome_tools.json Tool Type: Local Tools Count: 10

This page contains all tools defined in the alphagenome_tools.json configuration file.

Available Tools

AlphaGenome_atlas_list_scorers (Type: AlphaGenomeTool)

List every valid scorer name available from AlphaGenome Atlas (e.g. AVI_SCORE, RNA_SEQ, ATAC, CHI…

AlphaGenome_atlas_list_scorers tool specification

Tool Information:

• Name: AlphaGenome_atlas_list_scorers

• Type: AlphaGenomeTool

• Description: List every valid scorer name available from AlphaGenome Atlas (e.g. AVI_SCORE, RNA_SEQ, ATAC, CHIP_TF, …), with whether each is signed and how many tracks it has. Use this to discover valid values for the scorers parameter of AlphaGenome_atlas_lookup_variant/AlphaGenome_atlas_scan_interval instead of guessing. Hosted API via the alphagenome SDK; free for non-commercial use (set ALPHA_GENOME_API_KEY).

Parameters:

No parameters required.

Example Usage:

query = {
    "name": "AlphaGenome_atlas_list_scorers",
    "arguments": {
    }
}
result = tu.run(query)

AlphaGenome_atlas_lookup_variant (Type: AlphaGenomeTool)

Look up precomputed variant effect scores for a single-nucleotide substitution from AlphaGenome A…

AlphaGenome_atlas_lookup_variant tool specification

Tool Information:

• Name: AlphaGenome_atlas_lookup_variant

• Type: AlphaGenomeTool

• Description: Look up precomputed variant effect scores for a single-nucleotide substitution from AlphaGenome Atlas (Google DeepMind, released Sep 2026): a precomputed catalogue covering all ~9 billion possible human SNVs, including the unified AlphaGenome Variant Impact (AVI) score that combines AlphaGenome and AlphaMissense predictions into one coding+non-coding pathogenicity-style number. This is a database lookup (fast, high query rate), not a live model run – use AlphaGenome_score_variant instead for indels or live per-track scoring. Hosted API via the alphagenome SDK; free for non-commercial use (set ALPHA_GENOME_API_KEY).

Parameters:

• chromosome (string) (required) Chromosome, e.g. ‘chr22’.

• position (integer) (required) 1-based variant position.

• reference_bases (string) (required) Reference allele, e.g. ‘A’.

• alternate_bases (string) (required) Alternate allele, e.g. ‘C’.

• scorers ([‘array’, ‘null’]) (optional) Which precomputed Atlas scorers to return, e.g. [‘AVI_SCORE’, ‘RNA_SEQ’, ‘ATAC’, ‘CHIP_TF’]. Defaults to [‘AVI_SCORE’], the unified variant impact score.

• top_n ([‘integer’, ‘null’]) (optional) Number of top |effect| tracks to return per scorer (default 20).

Example Usage:

query = {
    "name": "AlphaGenome_atlas_lookup_variant",
    "arguments": {
        "chromosome": "example_value",
        "position": 10,
        "reference_bases": "example_value",
        "alternate_bases": "example_value"
    }
}
result = tu.run(query)

AlphaGenome_atlas_scan_interval (Type: AlphaGenomeTool)

Look up precomputed AlphaGenome Atlas scores for EVERY possible single-nucleotide substitution in…

AlphaGenome_atlas_scan_interval tool specification

Tool Information:

• Name: AlphaGenome_atlas_scan_interval

• Type: AlphaGenomeTool

• Description: Look up precomputed AlphaGenome Atlas scores for EVERY possible single-nucleotide substitution in a genomic region (<=10,000 bp), e.g. to rank all variants in a promoter by AVI score at once. This is a precomputed database read (fast, high query rate), not a live model run – for a single known variant use AlphaGenome_atlas_lookup_variant instead. Hosted API via the alphagenome SDK; free for non-commercial use (set ALPHA_GENOME_API_KEY).

Parameters:

• chromosome (string) (required) Chromosome, e.g. ‘chr22’.

• start (integer) (required) Interval start (0-based). Window (end - start) must be <= 10000 bp.

• end (integer) (required) Interval end. Window (end - start) must be <= 10000 bp.

• scorers ([‘array’, ‘null’]) (optional) Which precomputed Atlas scorers to return (see AlphaGenome_atlas_list_scorers), e.g. [‘AVI_SCORE’]. Defaults to [‘AVI_SCORE’].

• top_n ([‘integer’, ‘null’]) (optional) Number of top |effect| entries to return (default 20).

Example Usage:

query = {
    "name": "AlphaGenome_atlas_scan_interval",
    "arguments": {
        "chromosome": "example_value",
        "start": 10,
        "end": 10
    }
}
result = tu.run(query)

AlphaGenome_output_metadata (Type: AlphaGenomeTool)

Discover what DeepMind AlphaGenome can predict: per-modality track counts and a sample of valid o…

AlphaGenome_output_metadata tool specification

Tool Information:

• Name: AlphaGenome_output_metadata

• Type: AlphaGenomeTool

• Description: Discover what DeepMind AlphaGenome can predict: per-modality track counts and a sample of valid ontology_terms values (tissue/cell types) for each modality. Use this before calling AlphaGenome_score_variant/predict_variant/predict_interval/predict_sequence with an ontology_terms filter, to find valid values instead of guessing. Hosted API via the alphagenome SDK; free for non-commercial use (set ALPHA_GENOME_API_KEY).

Parameters:

• organism ([‘string’, ‘null’]) (optional) ‘human’ (default) or ‘mouse’.

Example Usage:

query = {
    "name": "AlphaGenome_output_metadata",
    "arguments": {
    }
}
result = tu.run(query)

AlphaGenome_predict_interval (Type: AlphaGenomeTool)

Predict multimodal genomic tracks for a genomic interval with DeepMind AlphaGenome (single DNA-se…

AlphaGenome_predict_interval tool specification

Tool Information:

• Name: AlphaGenome_predict_interval

• Type: AlphaGenomeTool

• Description: Predict multimodal genomic tracks for a genomic interval with DeepMind AlphaGenome (single DNA-sequence model; up to 1 Mb at single-base resolution). Returns a compact per-modality summary (track counts and shapes) for the requested outputs. Hosted API via the alphagenome SDK; free for non-commercial use (set ALPHA_GENOME_API_KEY).

Parameters:

• chromosome (string) (required) Chromosome, e.g. ‘chr19’.

• start (integer) (required) Interval start (0-based).

• end (integer) (required) Interval end.

• output_types ([‘array’, ‘null’]) (optional) Modalities to predict (default [‘RNA_SEQ’]): RNA_SEQ, ATAC, DNASE, CAGE, CHIP_HISTONE, CHIP_TF, SPLICE_SITES, SPLICE_SITE_USAGE, SPLICE_JUNCTIONS, CONTACT_MAPS, PROCAP.

• ontology_terms ([‘array’, ‘null’]) (optional) Optional tissue/cell ontology terms (e.g. [‘UBERON:0001114’] = liver) to restrict tracks.

• organism ([‘string’, ‘null’]) (optional) ‘human’ (default) or ‘mouse’.

• sequence_length ([‘string’, ‘null’]) (optional) Context window: 16KB, 100KB, 500KB, or 1MB (default).

Example Usage:

query = {
    "name": "AlphaGenome_predict_interval",
    "arguments": {
        "chromosome": "example_value",
        "start": 10,
        "end": 10
    }
}
result = tu.run(query)

AlphaGenome_predict_sequence (Type: AlphaGenomeTool)

Predict multimodal genomic tracks for a raw DNA sequence (no genome coordinates needed) with Deep…

AlphaGenome_predict_sequence tool specification

Tool Information:

• Name: AlphaGenome_predict_sequence

• Type: AlphaGenomeTool

• Description: Predict multimodal genomic tracks for a raw DNA sequence (no genome coordinates needed) with DeepMind AlphaGenome. Useful for synthetic constructs or sequences not anchored to a reference genome. The sequence length must match one of the model’s supported context windows (16384, 131072, 524288, or 1048576 bp; pad with ‘N’ as needed). Hosted API via the alphagenome SDK; free for non-commercial use (set ALPHA_GENOME_API_KEY).

Parameters:

• sequence (string) (required) Raw DNA sequence (A/C/G/T/N). Length must equal a supported context window (16384/131072/524288/1048576 bp).

• output_types ([‘array’, ‘null’]) (optional) Modalities to predict (default [‘RNA_SEQ’]): RNA_SEQ, ATAC, DNASE, CAGE, CHIP_HISTONE, CHIP_TF, SPLICE_SITES, SPLICE_SITE_USAGE, SPLICE_JUNCTIONS, CONTACT_MAPS, PROCAP.

• ontology_terms ([‘array’, ‘null’]) (optional) Optional tissue/cell ontology terms (e.g. [‘UBERON:0001114’] = liver) to restrict tracks. Use AlphaGenome_output_metadata to discover valid values.

• organism ([‘string’, ‘null’]) (optional) ‘human’ (default) or ‘mouse’.

Example Usage:

query = {
    "name": "AlphaGenome_predict_sequence",
    "arguments": {
        "sequence": "example_value"
    }
}
result = tu.run(query)

AlphaGenome_predict_variant (Type: AlphaGenomeTool)

Predict full multimodal tracks for BOTH the reference and alternate allele of a variant with Deep…

AlphaGenome_predict_variant tool specification

Tool Information:

• Name: AlphaGenome_predict_variant

• Type: AlphaGenomeTool

• Description: Predict full multimodal tracks for BOTH the reference and alternate allele of a variant with DeepMind AlphaGenome (distinct from AlphaGenome_score_variant, which reduces this to a single recommended per-gene effect score). Returns a compact per-modality summary (shape/track counts) for each allele – use this when you need the raw predicted tracks for both alleles rather than a scored comparison. Hosted API via the alphagenome SDK; free for non-commercial use (set ALPHA_GENOME_API_KEY).

Parameters:

• chromosome (string) (required) Chromosome, e.g. ‘chr22’.

• position (integer) (required) 1-based variant position.

• reference_bases (string) (required) Reference allele, e.g. ‘A’.

• alternate_bases (string) (required) Alternate allele, e.g. ‘C’.

• output_types ([‘array’, ‘null’]) (optional) Modalities to predict (default [‘RNA_SEQ’]): RNA_SEQ, ATAC, DNASE, CAGE, CHIP_HISTONE, CHIP_TF, SPLICE_SITES, SPLICE_SITE_USAGE, SPLICE_JUNCTIONS, CONTACT_MAPS, PROCAP.

• ontology_terms ([‘array’, ‘null’]) (optional) Optional tissue/cell ontology terms (e.g. [‘UBERON:0001114’] = liver) to restrict tracks. Use AlphaGenome_output_metadata to discover valid values.

• organism ([‘string’, ‘null’]) (optional) ‘human’ (default) or ‘mouse’.

• sequence_length ([‘string’, ‘null’]) (optional) Context window: 16KB, 100KB, 500KB, or 1MB (default).

Example Usage:

query = {
    "name": "AlphaGenome_predict_variant",
    "arguments": {
        "chromosome": "example_value",
        "position": 10,
        "reference_bases": "example_value",
        "alternate_bases": "example_value"
    }
}
result = tu.run(query)

AlphaGenome_score_interval (Type: AlphaGenomeTool)

Score the default gene-mask activity for a genomic interval with DeepMind AlphaGenome, without re…

AlphaGenome_score_interval tool specification

Tool Information:

• Name: AlphaGenome_score_interval

• Type: AlphaGenomeTool

• Description: Score the default gene-mask activity for a genomic interval with DeepMind AlphaGenome, without reference to a specific variant (e.g. baseline predicted expression level per nearby gene). Returns the top per-gene, per-track scores sorted by |effect|. For variant-effect scoring (ref vs alt), use AlphaGenome_score_variant instead. Hosted API via the alphagenome SDK; free for non-commercial use (set ALPHA_GENOME_API_KEY).

Parameters:

• chromosome (string) (required) Chromosome, e.g. ‘chr22’.

• start (integer) (required) Interval start (0-based).

• end (integer) (required) Interval end.

• organism ([‘string’, ‘null’]) (optional) ‘human’ (default) or ‘mouse’.

• sequence_length ([‘string’, ‘null’]) (optional) Context window: 16KB, 100KB, 500KB, or 1MB (default).

• top_n ([‘integer’, ‘null’]) (optional) Number of top |effect| entries to return (default 20).

Example Usage:

query = {
    "name": "AlphaGenome_score_interval",
    "arguments": {
        "chromosome": "example_value",
        "start": 10,
        "end": 10
    }
}
result = tu.run(query)

AlphaGenome_score_ism_variants (Type: AlphaGenomeTool)

In-silico saturation mutagenesis with DeepMind AlphaGenome: scores every possible single-nucleoti…

AlphaGenome_score_ism_variants tool specification

Tool Information:

• Name: AlphaGenome_score_ism_variants

• Type: AlphaGenomeTool

• Description: In-silico saturation mutagenesis with DeepMind AlphaGenome: scores every possible single-nucleotide substitution across a short window (<=500 bp) and ranks candidates by their single most extreme per-gene, per-track effect – a ‘which bases matter most’ scan for finding causal regulatory positions (e.g. within a promoter or enhancer). Each live call costs ~3x the window width in underlying model evaluations, so the window is capped at 500 bp; for a single known variant use AlphaGenome_score_variant instead. Hosted API via the alphagenome SDK; free for non-commercial use (set ALPHA_GENOME_API_KEY).

Parameters:

• chromosome (string) (required) Chromosome, e.g. ‘chr22’.

• start (integer) (required) ISM window start (0-based). Window (end - start) must be <= 500 bp.

• end (integer) (required) ISM window end. Window (end - start) must be <= 500 bp.

• output_type ([‘string’, ‘null’]) (optional) Modality/scorer to rank by, i.e. a key of alphagenome.models.variant_scorers.RECOMMENDED_VARIANT_SCORERS. RNA_SEQ (default), ATAC, DNASE, CAGE, CHIP_HISTONE, CHIP_TF, PROCAP, SPLICE_SITES, SPLICE_SITE_USAGE, SPLICE_JUNCTIONS, POLYADENYLATION, CONTACT_MAPS, or an _ACTIVE variant of ATAC/DNASE/CAGE/CHIP_TF/CHIP_HISTONE/PROCAP/RNA_SEQ.

• organism ([‘string’, ‘null’]) (optional) ‘human’ (default) or ‘mouse’.

• sequence_length ([‘string’, ‘null’]) (optional) Context window: 16KB, 100KB, 500KB, or 1MB (default).

• top_n ([‘integer’, ‘null’]) (optional) Number of top |effect| entries to return (default 20).

Example Usage:

query = {
    "name": "AlphaGenome_score_ism_variants",
    "arguments": {
        "chromosome": "example_value",
        "start": 10,
        "end": 10
    }
}
result = tu.run(query)

AlphaGenome_score_variant (Type: AlphaGenomeTool)

Score a regulatory variant’s effect with DeepMind AlphaGenome (Avsec, Nature 2026), the hosted su…

AlphaGenome_score_variant tool specification

Tool Information:

• Name: AlphaGenome_score_variant

• Type: AlphaGenomeTool

• Description: Score a regulatory variant’s effect with DeepMind AlphaGenome (Avsec, Nature 2026), the hosted successor to Enformer/Borzoi. Predicts ref vs alt over up to 1 Mb at single-base resolution and returns the recommended per-track effect scores (sorted by |effect|). Hosted API via the alphagenome SDK; free for non-commercial use (set ALPHA_GENOME_API_KEY).

Parameters:

• chromosome (string) (required) Chromosome, e.g. ‘chr22’.

• position (integer) (required) 1-based variant position.

• reference_bases (string) (required) Reference allele, e.g. ‘A’.

• alternate_bases (string) (required) Alternate allele, e.g. ‘C’.

• output_type ([‘string’, ‘null’]) (optional) Modality to score, i.e. a key of alphagenome.models.variant_scorers.RECOMMENDED_VARIANT_SCORERS. RNA_SEQ (default), ATAC, DNASE, CAGE, CHIP_HISTONE, CHIP_TF, PROCAP, SPLICE_SITES, SPLICE_SITE_USAGE, SPLICE_JUNCTIONS, POLYADENYLATION, CONTACT_MAPS, or an _ACTIVE variant of ATAC/DNASE/CAGE/CHIP_TF/CHIP_HISTONE/PROCAP/RNA_SEQ.

• organism ([‘string’, ‘null’]) (optional) ‘human’ (default) or ‘mouse’.

• sequence_length ([‘string’, ‘null’]) (optional) Context window: 16KB, 100KB, 500KB, or 1MB (default).

• top_n ([‘integer’, ‘null’]) (optional) Number of top |effect| tracks to return (default 20).

Example Usage:

query = {
    "name": "AlphaGenome_score_variant",
    "arguments": {
        "chromosome": "example_value",
        "position": 10,
        "reference_bases": "example_value",
        "alternate_bases": "example_value"
    }
}
result = tu.run(query)

Navigation

• Available Tools - Back to Tools Overview

• Tool Loading Tutorial - Loading Local Tools