How do I use igvf-portal-mcp?

1. Click on "Install Server". 2. Wait a few minutes for the server to deploy. Once ready, it will show a "Started" state. 3. In the chat, type @ followed by the MCP server name and your instructions, e.g., "@igvf-portal-mcp search for gene TP53" That's it! The server will respond to your query, and you can continue using it as needed. Here is a step-by-step guide with screenshots.

igvf-portal-mcp

Official

by IGVF-DACC

Overview Schema Related Servers Score Discussions

Python

Remote

IGVF Portal MCP Server

An MCP server that exposes the IGVF Data Portal API as tools for AI coding agents. Use it to search, filter, download files, and generate reports from the portal directly within an agent session.

The server is built on top of the IGVF Python client, which documents the full underlying API.

Related MCP server: Omics AI MCP Server

Examples (using Claude Code)

List all item types

❯ What types exist on the IGVF portal?

● igvf-portal-mcp - igvf_portal_list_item_types (MCP)
  ⎿  {
       "result": "[\n  \"AccessKey\",\n  \"AlignmentFile\",\n  \"AnalysisSet\",\n  \"AnalysisStep\",\n  \"AnalysisStepVersion\",\n  \"AssayTerm\",\n  \"AuxiliarySet\",\n  \"Award\",\n  \"Biomarker\",\n  \"ConfigurationFile\",
     \n  \"ConstructLibrarySet\",\n  \"CrisprModification\",\n  \"CuratedSet\",\n  \"DegronModification\",\n  \"Document\",\n  \"Gene\",\n  \"HumanDonor\",\n  \"Image\",\n  \"ImageFile\",\n  \"InVitroSystem\",\n  \"IndexFile\
     … +5 lines (ctrl+o to expand)

● The IGVF portal has 57 item types:

  Files
  - AlignmentFile, ConfigurationFile, ImageFile, IndexFile, MatrixFile, ModelFile, ReferenceFile, SequenceFile, SignalFile, TabularFile

  Sets / Collections
  - AnalysisSet, AuxiliarySet, ConstructLibrarySet, CuratedSet, ModelSet, MeasurementSet, PredictionSet

  Samples / Biospecimens
  - InVitroSystem, MultiplexedSample, PrimaryCell, TechnicalSample, Tissue, WholeOrganism

  Donors
  - HumanDonor, RodentDonor

  Genomics / Biology
  - Gene, OpenReadingFrame, Biomarker, CrisprModification, DegronModification

  Terms / Ontologies
  - AssayTerm, PhenotypeTerm, PlatformTerm, SampleTerm, PhenotypicFeature

  Quality Metrics
  - MpraQualityMetric, PerturbSeqQualityMetric, SingleCellAtacSeqQualityMetric, SingleCellRnaSeqQualityMetric, StarrSeqQualityMetric

  Analysis / Workflows
  - AnalysisStep, AnalysisStepVersion, Software, SoftwareVersion, Workflow

  Administrative / Metadata
  - AccessKey, Award, Document, Image, InstitutionalCertificate, Lab, Page, Publication, Source, Treatment, User

Compare item types using schemas

❯ What's the difference between measurement set and analysis set?

● igvf-portal-mcp - igvf_portal_get_schema (MCP)(item_type: "MeasurementSet")
  ⎿  ⚠ Large MCP response (~15.3k tokens), this can fill up context quickly
  ⎿  {"result":"{\n  \"title\": \"Measurement Set\",\n  \"$id\": \"/profiles/measurement_set.json\", ...
     … +276 lines (ctrl+o to expand)

● igvf-portal-mcp - igvf_portal_get_schema (MCP)(item_type: "AnalysisSet")
  ⎿  {"result":"{\n  \"title\": \"Analysis Set\",\n  \"$id\": \"/profiles/analysis_set.json\", ...
     … +142 lines (ctrl+o to expand)

● Here's the key distinction, straight from the schema descriptions:

  MeasurementSet: Hosts raw data files (e.g. FASTQs) resulting from sequencing of a library
  prepared from the nucleic acids of the sample. The sample-specific raw sequencing results
  are captured here.

  AnalysisSet: Represents the results of a computational analysis of raw genomic data or
  other analyses.

  ┌──────────────────┬─────────────────────────────────────────────┬─────────────────────────────────────────────────┐
  │                  │               MeasurementSet                │                   AnalysisSet                   │
  ├──────────────────┼─────────────────────────────────────────────┼─────────────────────────────────────────────────┤
  │ Purpose          │ Raw experimental data                       │ Computational analysis results                  │
  ├──────────────────┼─────────────────────────────────────────────┼─────────────────────────────────────────────────┤
  │ Typical files    │ FASTQs, raw reads                           │ BAMs, matrices, BEDs, TSVs                      │
  ├──────────────────┼─────────────────────────────────────────────┼─────────────────────────────────────────────────┤
  │ Input            │ Biological samples                          │ Other file sets (MeasurementSets, etc.)         │
  ├──────────────────┼─────────────────────────────────────────────┼─────────────────────────────────────────────────┤
  │ Required fields  │ assay_term, samples, preferred_assay_titles │ file_set_type                                   │
  ├──────────────────┼─────────────────────────────────────────────┼─────────────────────────────────────────────────┤
  │ Types            │ (single type)                               │ intermediate analysis, principal analysis       │
  ├──────────────────┼─────────────────────────────────────────────┼─────────────────────────────────────────────────┤
  │ Key relationship │ Feeds into AnalysisSets                     │ Has input_file_sets pointing to MeasurementSets │
  └──────────────────┴─────────────────────────────────────────────┴─────────────────────────────────────────────────┘

  A typical flow is: MeasurementSet (raw FASTQs) → AnalysisSet (intermediate)
  (e.g. BAM + fragment file) → AnalysisSet (principal) (final matrices, quantifications).

Search for genes by name

❯ Search IGVF portal for ep300 genes

● igvf-portal-mcp - igvf_portal_search (MCP)(query: "ep300", type: ["Gene"])
  ⎿  {"result":"{\n  \"total\": 11,\n  \"returned\": 11, ...

● Found 11 genes related to EP300:

  ┌───────────┬───────────────────────────────────────────────────┬──────────────┬────────────────────┬─────────────────────────────┐
  │  Symbol   │                       Name                        │     Taxa     │      Gene ID       │          Location           │
  ├───────────┼───────────────────────────────────────────────────┼──────────────┼────────────────────┼─────────────────────────────┤
  │ EP300     │ E1A binding protein p300                          │ Homo sapiens │ ENSG00000100393    │ chr22:41,092,591–41,180,077 │
  ├───────────┼───────────────────────────────────────────────────┼──────────────┼────────────────────┼─────────────────────────────┤
  │ EP300-AS1 │ EP300 antisense RNA 1                             │ Homo sapiens │ ENSG00000231993    │ chr22:41,174,590–41,197,508 │
  ├───────────┼───────────────────────────────────────────────────┼──────────────┼────────────────────┼─────────────────────────────┤
  │ Ep300     │ E1A binding protein p300                          │ Mus musculus │ ENSMUSG00000055024 │ chr15:81,469,551–81,536,278 │
  ├───────────┼───────────────────────────────────────────────────┼──────────────┼────────────────────┼─────────────────────────────┤
  │ EID1      │ EP300 interacting inhibitor of differentiation 1  │ Homo sapiens │ ENSG00000255302    │ chr15                       │
  ├───────────┼───────────────────────────────────────────────────┼──────────────┼────────────────────┼─────────────────────────────┤
  │ EID2      │ EP300 interacting inhibitor of differentiation 2  │ Homo sapiens │ ENSG00000176396    │ chr19                       │
  ├───────────┼───────────────────────────────────────────────────┼──────────────┼────────────────────┼─────────────────────────────┤
  │ EID2B     │ EP300 interacting inhibitor of differentiation 2B │ Homo sapiens │ ENSG00000176401    │ chr19                       │
  ├───────────┼───────────────────────────────────────────────────┼──────────────┼────────────────────┼─────────────────────────────┤
  │ EID3      │ EP300 interacting inhibitor of differentiation 3  │ Homo sapiens │ ENSG00000255150    │ chr12                       │
  ├───────────┼───────────────────────────────────────────────────┼──────────────┼────────────────────┼─────────────────────────────┤
  │ Eid1      │ EP300 interacting inhibitor of differentiation 1  │ Mus musculus │ ENSMUSG00000091337 │ chr2                        │
  ├───────────┼───────────────────────────────────────────────────┼──────────────┼────────────────────┼─────────────────────────────┤
  │ Eid2      │ EP300 interacting inhibitor of differentiation 2  │ Mus musculus │ ENSMUSG00000046058 │ chr7                        │
  ├───────────┼───────────────────────────────────────────────────┼──────────────┼────────────────────┼─────────────────────────────┤
  │ Eid2b     │ EP300 interacting inhibitor of differentiation 2B │ Mus musculus │ ENSMUSG00000070705 │ chr7                        │
  ├───────────┼───────────────────────────────────────────────────┼──────────────┼────────────────────┼─────────────────────────────┤
  │ Eid3      │ EP300 interacting inhibitor of differentiation 3  │ Mus musculus │ ENSMUSG00000109864 │ chr10                       │
  └───────────┴───────────────────────────────────────────────────┴──────────────┴────────────────────┴─────────────────────────────┘

  The core EP300 gene itself appears in both human (ENSG00000100393) and mouse
  (ENSMUSG00000055024) versions, along with its antisense RNA and several
  EP300-interacting inhibitor of differentiation (EID) family members.

Look up an assay term

❯ What is assay term OBI:0002762?

● igvf-portal-mcp - igvf_portal_search (MCP)(query: "OBI:0002762", type: ["AssayTerm"])
  ⎿  {
       "result": "{\"total\": 1, \"returned\": 1, \"results\": [{\"preferred_assay_titles\": [\"SHARE-seq\",
         \"10x multiome\", \"10x multiome with MULTI-seq\", \"miDOGMA-seq\", \"scMultiome-NT-seq\",
         \"10x snATAC-seq with Scale pre-indexing\"], \"term_id\": \"OBI:0002762\",
         \"term_name\": \"single-nucleus ATAC-seq\", ...
     … +10 lines (ctrl+o to expand)

● OBI:0002762 is single-nucleus ATAC-seq (snATAC-seq).

  - Definition: An ATAC-seq assay in which single nuclei are isolated from frozen tissue samples,
    through a protocol designed to maximally preserve nucleus integrity during sample processing
    and optimize transposase-mediated fragmentation of chromatin in individual nuclei.
  - Assay slim: chromatin accessibility
  - Preferred assay titles on IGVF: SHARE-seq, 10x multiome, 10x multiome with MULTI-seq,
    miDOGMA-seq, scMultiome-NT-seq, 10x snATAC-seq with Scale pre-indexing

Get a facet breakdown

❯ Give me a breakdown of measurement sets by preferred assay title

● igvf-portal-mcp - igvf_portal_facets (MCP)(type: ["MeasurementSet"])
  ⎿  Error: result (120,385 characters) exceeds maximum allowed tokens. Output has been saved to ...
     Format: JSON with schema: {result: string}
     … +6 lines (ctrl+o to expand)

● Bash(cat .../tool-results/mcp-igvf-portal-mcp-igvf_portal_facets-....txt | python3 -c "...")
  ⎿  Total MeasurementSets: 5129

       1470  10x multiome
     … +39 lines (ctrl+o to expand)

● MeasurementSets by preferred assay title (total: 5,129):

  ┌───────┬───────────────────────────────────┐
  │ Count │       Preferred Assay Title       │
  ├───────┼───────────────────────────────────┤
  │ 1,470 │ 10x multiome                      │
  ├───────┼───────────────────────────────────┤
  │   798 │ SGE                               │
  ├───────┼───────────────────────────────────┤
  │   519 │ SHARE-seq                         │
  ├───────┼───────────────────────────────────┤
  │   488 │ Variant-EFFECTS                   │
  ├───────┼───────────────────────────────────┤
  │   333 │ Parse SPLiT-seq                   │
  ├───────┼───────────────────────────────────┤
  │   268 │ CRISPR FlowFISH screen            │
  ├───────┼───────────────────────────────────┤
  │   196 │ Perturb-seq                       │
  ├───────┼───────────────────────────────────┤
  │   144 │ VAMP-seq (MultiSTEP)              │
  ├───────┼───────────────────────────────────┤
  │   118 │ 10x multiome with MULTI-seq       │
  ├───────┼───────────────────────────────────┤
  │   108 │ CRISPR FACS screen                │
  ├───────┼───────────────────────────────────┤
  │    95 │ RNA-seq                           │
  ├───────┼───────────────────────────────────┤
  │    76 │ mtscMultiome                      │
  ├───────┼───────────────────────────────────┤
  │    54 │ MPRA                              │
  ├───────┼───────────────────────────────────┤
  │    48 │ TAP-seq                           │
  ├───────┼───────────────────────────────────┤
  │    48 │ scNT-seq2                         │
  ├───────┼───────────────────────────────────┤
  │    41 │ ATAC-seq                          │
  ├───────┼───────────────────────────────────┤
  │    40 │ Proliferation CRISPR screen       │
  ├───────┼───────────────────────────────────┤
  │    36 │ VAMP-seq                          │
  ├───────┼───────────────────────────────────┤
  │    36 │ snM3C-seq                         │
  ├───────┼───────────────────────────────────┤
  │    34 │ snMCT-seq                         │
  ├───────┼───────────────────────────────────┤
  │    27 │ lentiMPRA                         │
  ├───────┼───────────────────────────────────┤
  │    23 │ WGS                               │
  ├───────┼───────────────────────────────────┤
  │    20 │ HiCAR                             │
  ├───────┼───────────────────────────────────┤
  │    19 │ Parse Perturb-seq                 │
  ├───────┼───────────────────────────────────┤
  │    15 │ STARR-seq                         │
  ├───────┼───────────────────────────────────┤
  │     8 │ ONT Fiber-seq                     │
  ├───────┼───────────────────────────────────┤
  │     8 │ ONT dRNA                          │
  ├───────┼───────────────────────────────────┤
  │     8 │ ONT direct WGS                    │
  ├───────┼───────────────────────────────────┤
  │     8 │ scCRISPR screen                   │
  ├───────┼───────────────────────────────────┤
  │     6 │ MPRA (scQer)                      │
  ├───────┼───────────────────────────────────┤
  │     6 │ Migration CRISPR screen           │
  ├───────┼───────────────────────────────────┤
  │     6 │ electroporated MPRA               │
  ├───────┼───────────────────────────────────┤
  │     4 │ Arrayed semi-qY2H v1/v2/v3        │
  ├───────┼───────────────────────────────────┤
  │     4 │ scATAC-seq                        │
  ├───────┼───────────────────────────────────┤
  │     4 │ scNT-seq                          │
  ├───────┼───────────────────────────────────┤
  │     3 │ scRNA-seq                         │
  ├───────┼───────────────────────────────────┤
  │     1 │ DUAL-IPA                          │
  ├───────┼───────────────────────────────────┤
  │     1 │ Variant painting via fluorescence │
  └───────┴───────────────────────────────────┘

  10x multiome dominates at 1,470 (29%), followed by SGE (798) and SHARE-seq (519).

Discover filter parameters

❯ What can I filter on for analysis sets?

● igvf-portal-mcp - igvf_portal_get_endpoint_params (MCP)(endpoint: "analysis_sets")
  ⎿  {"result":"{\n  \"endpoint\": \"analysis_sets\",\n  \"standard_params\": [\n    \"limit\",\n    \"query\",\n    \"sort\"\n  ],\n  \"filter_params\": [\n    {\n      \"collection_param\": \"id\",\n      \"search_field\":
     \"@id\",\n      \"type\": \"list[str]\"\n    },\n    {\n      \"collection_param\": \"accession\",\n      \"search_field\": \"accession\",\n      \"type\": \"list[str]\"\n    },\n    {\n      \"collection_param\": \"alia
     ses\",\n      \"search_field\": \"aliases\",\n      \"type\": \"list[str]\"\n    },\n    {\n      \"collection_param\": \"alternate_accessions\",\n      \"search_field\": \"alternate_accessions\",\n      \"type\": \"list
     … +119 lines (ctrl+o to expand)

● Here are all filterable fields for AnalysisSet, grouped by category:

  Identity: @id, accession, uuid, aliases, alternate_accessions, doi, status,
    controlled_access, data_use_limitation_summaries, is_on_anvil

  Assay: assay_titles, preferred_assay_titles, file_set_type,
    uniform_pipeline_status, enrichment_designs,
    functional_assay_mechanisms.term_name

  Samples: samples.@id, samples.accession, samples.summary, sample_summary,
    samples.classifications, samples.taxa, samples.sample_terms.term_name,
    samples.disease_terms.term_name, samples.targeted_sample_term.term_name,
    samples.modifications.modality, samples.treatments.treatment_term_name,
    samples.treatments.treatment_type, samples.treatments.purpose,
    samples.growth_medium, samples.cellular_sub_pool,
    samples.institutional_certificates.data_use_limitation

  Donors: donors.@id, donors.accession, donors.taxa, donors.sex,
    donors.strain_background

  Files: files.@id, files.accession, files.content_type, files.file_format,
    files.file_format_type, files.assembly, files.transcriptome_annotation,
    files.file_size, files.status, files.upload_status,
    files.controlled_access, files.sequencing_platform.term_name,
    files.cell_type_annotation.term_name

  Input / linked sets: input_file_sets.@id, input_file_sets.accession,
    input_file_sets.file_set_type, input_for, construct_library_sets.@id,
    construct_library_sets.file_set_type,
    construct_library_sets.selection_criteria,
    construct_library_sets.small_scale_gene_list.symbol,
    construct_library_sets.associated_phenotypes.term_name,
    targeted_genes.symbol, demultiplexed_samples

  Workflows / pipeline: workflows.@id, workflows.accession, workflows.name,
    workflows.uniform_pipeline, workflows.workflow_version, pipeline_parameters

  Lab / award: lab.@id, lab.title, award.@id, award.component, award.title,
    award.contact_pi.title, publications.publication_identifiers

  Timestamps: creation_timestamp, release_timestamp,
    submitted_files_timestamp, preview_timestamp

Search with field filters

❯ Give me an example 10x multiome analysis set

● igvf-portal-mcp - igvf_portal_search (MCP)(type: ["AnalysisSet"], field_filters: {"preferred_assay_titles":"10x multiome"}, limit: "1")
  ⎿  {
       "result": "{\n  \"total\": 726,\n  \"returned\": 1,\n  \"results\": [\n    {\n      \"doi\": \"10.65695/IGVFDS9564JXGL\",\n      \"input_file_sets\": [\n        \"/measurement-sets/IGVFDS2443ZQHK/\",\n        \"/measur
     ement-sets/IGVFDS9885KGYZ/\"\n      ],\n      \"release_timestamp\": \"2025-11-12T18:17:10.823135+00:00\",\n      \"lab\": \"/labs/ryan-corces/\",\n      \"award\": \"/awards/HG012076/\",\n      \"accession\": \"IGVFDS95
     … +11 lines (ctrl+o to expand)

● Here's an example 10x multiome AnalysisSet:

  IGVFDS9564JXGL — https://data.igvf.org/analysis-sets/IGVFDS9564JXGL/

  ┌───────────────────┬──────────────────────────────────────────────────────────────────────────────────────────┐
  │       Field       │                                          Value                                           │
  ├───────────────────┼──────────────────────────────────────────────────────────────────────────────────────────┤
  │ Type              │ intermediate analysis                                                                    │
  ├───────────────────┼──────────────────────────────────────────────────────────────────────────────────────────┤
  │ Status            │ released                                                                                 │
  ├───────────────────┼──────────────────────────────────────────────────────────────────────────────────────────┤
  │ Lab               │ Ryan Corces                                                                              │
  ├───────────────────┼──────────────────────────────────────────────────────────────────────────────────────────┤
  │ Assays            │ single-nucleus ATAC-seq + single-nucleus RNA-seq (10x multiome, genetically multiplexed) │
  ├───────────────────┼──────────────────────────────────────────────────────────────────────────────────────────┤
  │ Sample            │ Homo sapiens multiplexed sample — posterior cingulate gyrus, putamen, substantia nigra   │
  ├───────────────────┼──────────────────────────────────────────────────────────────────────────────────────────┤
  │ Donors            │ 8 human donors                                                                           │
  ├───────────────────┼──────────────────────────────────────────────────────────────────────────────────────────┤
  │ Pipeline          │ uniform pipeline (completed)                                                             │
  ├───────────────────┼──────────────────────────────────────────────────────────────────────────────────────────┤
  │ Controlled access │ No                                                                                       │
  └───────────────────┴──────────────────────────────────────────────────────────────────────────────────────────┘

  Files (6):
  - IGVFFI3882BXBL — alignment file
  - IGVFFI6307SAFZ, IGVFFI9552YLAE — index files
  - IGVFFI6617CCIB, IGVFFI8348ZDXP — matrix files
  - IGVFFI0795NBVQ — tabular file

  Input MeasurementSets: IGVFDS2443ZQHK, IGVFDS9885KGYZ

Look up files in an analysis set

❯ What files are in /analysis-sets/IGVFDS9564JXGL/?

● igvf-portal-mcp - igvf_portal_get_by_id (MCP)(resource_id: "/analysis-sets/IGVFDS9564JXGL/")
  ⎿  {
       "result": "{\n  \"doi\": \"10.65695/IGVFDS9564JXGL\",\n  \"input_file_sets\": [\n    \"/measurement-sets/IGVFDS2443ZQHK/\", ...
     … +9 lines (ctrl+o to expand)

● IGVFDS9564JXGL is an intermediate analysis set from the Ryan Corces lab (10x multiome —
  snATAC-seq + snRNA-seq, genetically multiplexed). It contains 6 files:

  ┌────────────────┬────────────────┐
  │   Accession    │      Type      │
  ├────────────────┼────────────────┤
  │ IGVFFI3882BXBL │ Alignment file │
  ├────────────────┼────────────────┤
  │ IGVFFI6307SAFZ │ Index file     │
  ├────────────────┼────────────────┤
  │ IGVFFI9552YLAE │ Index file     │
  ├────────────────┼────────────────┤
  │ IGVFFI6617CCIB │ Matrix file    │
  ├────────────────┼────────────────┤
  │ IGVFFI8348ZDXP │ Matrix file    │
  ├────────────────┼────────────────┤
  │ IGVFFI0795NBVQ │ Tabular file   │
  └────────────────┴────────────────┘

  The analysis covers a multiplexed sample from 8 human donors across three brain
  regions: posterior cingulate gyrus, putamen, and substantia nigra.

Drill into individual files

❯ Give me details on the BED and BAM files.

● igvf-portal-mcp - igvf_portal_get_by_id (MCP)(resource_id: "IGVFFI3882BXBL")
  ⎿  {
       "result": "{\n  \"controlled_access\": false,\n  \"release_timestamp\": \"2025-11-12T18:17:11.471572+00:00\", ...
     … +11 lines (ctrl+o to expand)

● igvf-portal-mcp - igvf_portal_get_by_id (MCP)(resource_id: "IGVFFI0795NBVQ")
  ⎿  {
       "result": "{\n  \"controlled_access\": false,\n  \"assembly\": \"GRCh38\",\n  \"file_format_type\": \"bed3+\", ...
     … +11 lines (ctrl+o to expand)

● BAM — IGVFFI3882BXBL
  - Format: BAM (unfiltered alignments), assembly GRCh38
  - Content: Raw aligned reads from Chromap (snATAC-seq)
  - Size: 19.4 GB
  - Reads: 454,907,788
  - Derived from: 9 sequence files + 3 configuration files
  - S3: s3://igvf-public/2025/10/21/ec04ba8e.../IGVFFI3882BXBL.bam

  BED — IGVFFI0795NBVQ
  - Format: BED (bed3+, gzipped), assembly GRCh38
  - Content: Raw ATAC-seq fragment file from Chromap (unfiltered)
  - Size: 2.18 GB
  - Derived from: same 9 sequence files + 3 configuration files as the BAM
  - S3: s3://igvf-public/2025/10/21/b7d07a7c.../IGVFFI0795NBVQ.bed.gz

  Both were produced by the IGVF DACC processing pipeline using the same workflow
  (IGVFWF6403DVII) and are unfiltered outputs. The BED is used as input to produce
  the index file IGVFFI6307SAFZ, and the BAM feeds into index file IGVFFI9552YLAE.

Explore the samples

❯ What samples were used?

● igvf-portal-mcp - igvf_portal_get_by_id (MCP)(resource_id: "/multiplexed-samples/IGVFSM6670KYMT/")
  ⎿  {
       "result": "{\n  \"preview_timestamp\": \"2025-03-25T22:41:47.611954+00:00\",\n  \"release_timestamp\": \"2025-09-17T18:47:59.370110+00:00\", ...
     … +9 lines (ctrl+o to expand)

● The multiplexed sample consists of 8 tissue samples from 8 European donors, all postmortem
  brain tissue sourced from Banner Health:

  ┌────────────────┬───────────────────────────┬───────────┬─────┬──────────────────────────────────────────────────┐
  │   Accession    │          Tissue           │ Donor sex │ Age │                     Disease                      │
  ├────────────────┼───────────────────────────┼───────────┼─────┼──────────────────────────────────────────────────┤
  │ IGVFSM8850QXUW │ Putamen                   │ Male      │ 85  │ Parkinson's disease                              │
  ├────────────────┼───────────────────────────┼───────────┼─────┼──────────────────────────────────────────────────┤
  │ IGVFSM9308VHFW │ Putamen                   │ Female    │ 82  │ Parkinson's disease                              │
  ├────────────────┼───────────────────────────┼───────────┼─────┼──────────────────────────────────────────────────┤
  │ IGVFSM9550JLTC │ Putamen                   │ Male      │ 78  │ Parkinson's disease                              │
  ├────────────────┼───────────────────────────┼───────────┼─────┼──────────────────────────────────────────────────┤
  │ IGVFSM9314PSQK │ Substantia nigra          │ Male      │ 74  │ Parkinson's disease, cerebral amyloid angiopathy │
  ├────────────────┼───────────────────────────┼───────────┼─────┼──────────────────────────────────────────────────┤
  │ IGVFSM3597NWGE │ Substantia nigra          │ Female    │ 70  │ Parkinson's disease                              │
  ├────────────────┼───────────────────────────┼───────────┼─────┼──────────────────────────────────────────────────┤
  │ IGVFSM0019LQKQ │ Substantia nigra          │ Male      │ 70  │ Parkinson's disease                              │
  ├────────────────┼───────────────────────────┼───────────┼─────┼──────────────────────────────────────────────────┤
  │ IGVFSM7450JGGC │ Substantia nigra          │ Male      │ 79  │ None listed                                      │
  ├────────────────┼───────────────────────────┼───────────┼─────┼──────────────────────────────────────────────────┤
  │ IGVFSM4995KUZR │ Posterior cingulate gyrus │ Male      │ 83  │ Parkinson's disease, cerebral amyloid angiopathy │
  └────────────────┴───────────────────────────┴───────────┴─────┴──────────────────────────────────────────────────┘

  7 of 8 donors have Parkinson's disease; 2 also have cerebral amyloid angiopathy. Ages
  range from 70–85. The one donor without a listed disease (IGVFSM7450JGGC) likely
  serves as a control.

Download and preview a file

❯ Download IGVFFI7253BTIQ and give me a preview

● Let me first get info on this file, then download it.

● igvf-portal-mcp - igvf_portal_get_by_id (MCP)(resource_id: "IGVFFI7253BTIQ")
  ⎿  {
       "result": "{\"controlled_access\": false, \"assembly\": \"GRCh38\", \"transcriptome_annotation\": \"GENCODE 43\",
         \"filtered\": true, \"accession\": \"IGVFFI7253BTIQ\", \"status\": \"released\", \"file_format\": \"tsv\",
         \"file_size\": 50518, \"content_type\": \"differential gene expression quantifications\", ...
     … +10 lines (ctrl+o to expand)

● It's a TSV file (~50 KB). Downloading now.

● igvf-portal-mcp - igvf_portal_download (MCP)(file_id: "IGVFFI7253BTIQ", save_path: "/tmp/IGVFFI7253BTIQ.tsv.gz")
  ⎿  {"saved_to": "/tmp/IGVFFI7253BTIQ.tsv.gz", "bytes": 50518}

● Here's a preview of IGVFFI7253BTIQ:

  File: differential gene expression quantifications (TSV, GRCh38 GENCODE 43, Parse Perturb-seq)
  Description: Analysis results for direct targeting (filtered)
  Rows: 969 data rows + 1 header

  Columns:

  ┌─────────────────────────┬─────────────────────────────────────────┐
  │         Column          │               Description               │
  ├─────────────────────────┼─────────────────────────────────────────┤
  │ intended_target_name    │ CRISPR target gene name                 │
  ├─────────────────────────┼─────────────────────────────────────────┤
  │ guide_id(s)             │ Guide RNA identifiers (comma-separated) │
  ├─────────────────────────┼─────────────────────────────────────────┤
  │ targeting_chr/start/end │ Genomic coordinates of the target       │
  ├─────────────────────────┼─────────────────────────────────────────┤
  │ gene_id                 │ Ensembl gene ID (the tested gene)       │
  ├─────────────────────────┼─────────────────────────────────────────┤
  │ gene_symbol             │ Gene symbol                             │
  ├─────────────────────────┼─────────────────────────────────────────┤
  │ sceptre_log2_fc         │ Log2 fold change (SCEPTRE method)       │
  ├─────────────────────────┼─────────────────────────────────────────┤
  │ sceptre_p_value         │ Raw p-value                             │
  ├─────────────────────────┼─────────────────────────────────────────┤
  │ sceptre_adj_p_value     │ Adjusted p-value                        │
  ├─────────────────────────┼─────────────────────────────────────────┤
  │ significant             │ TRUE/FALSE significance call            │
  ├─────────────────────────┼─────────────────────────────────────────┤
  │ type                    │ Direct_Targeting for all rows           │
  └─────────────────────────┴─────────────────────────────────────────┘

  Sample rows:

  ┌─────────────────┬────────┬─────────┬──────────┬─────────────┐
  │ intended_target │  gene  │ log2_fc │  adj_p   │ significant │
  ├─────────────────┼────────┼─────────┼──────────┼─────────────┤
  │ BAD             │ BAD    │ 0       │ 1        │ FALSE       │
  ├─────────────────┼────────┼─────────┼──────────┼─────────────┤
  │ TFPI            │ TFPI   │ -0.876  │ 1.52e-9  │ TRUE        │
  ├─────────────────┼────────┼─────────┼──────────┼─────────────┤
  │ SLC7A2          │ SLC7A2 │ -1.927  │ 1.57e-18 │ TRUE        │
  ├─────────────────┼────────┼─────────┼──────────┼─────────────┤
  │ HSPB6           │ HSPB6  │ -2.223  │ 1.06e-6  │ TRUE        │
  ├─────────────────┼────────┼─────────┼──────────┼─────────────┤
  │ CSDE1           │ CSDE1  │ -2.317  │ 5.60e-34 │ TRUE        │
  ├─────────────────┼────────┼─────────┼──────────┼─────────────┤
  │ MGST1           │ MGST1  │ -2.333  │ 1.98e-21 │ TRUE        │
  └─────────────────┴────────┴─────────┴──────────┴─────────────┘

  Most perturbations show negative log2 FC (gene knockdown via CRISPRi/dCas9-KRAB),
  with SCEPTRE used for differential expression testing.

Setup

Requirements

Python ≥ 3.11
uv

Dependencies are listed in requirements.txt and managed automatically by uv.

Run

uv run server.py

Connect to an AI coding agent

Add the server to your agent's MCP configuration. For example, in Claude Code (.mcp.json):

{
  "mcpServers": {
    "igvf-portal": {
      "command": "uv",
      "args": ["run", "server.py"]
    }
  }
}

A .mcp.json with this configuration is included in this repo.

Environment variables

Variable	Required	Description
`IGVF_ACCESS_KEY`	No	API key for authenticated access (unreleased data)
`IGVF_SECRET_ACCESS_KEY`	No	API secret paired with `IGVF_ACCESS_KEY`
`IGVF_HOST`	No	Portal host override, e.g. `https://api.sandbox.igvf.org`
`HTTPS_PROXY` / `HTTP_PROXY`	No	Proxy server URL
`REQUESTS_CA_BUNDLE` / `SSL_CERT_FILE`	No	Custom CA certificate path

Without credentials, the server connects anonymously to the production portal.

Tool Reference

All tools are prefixed igvf_portal_.

Primary Tools

Tool	Description
`igvf_portal_get_by_id`	Retrieve a single item by `@id`, accession, or UUID
`igvf_portal_search`	Search the portal with free text and/or field filters
`igvf_portal_get_collection`	List items from a collection endpoint
`igvf_portal_download`	Download a single file to a local path
`igvf_portal_report`	Generate a TSV report for an item type
`igvf_portal_batch_download`	Get download URLs for files in matching file sets

Supporting Tools

Tool	Description
`igvf_portal_get_schema`	Return the JSON schema for an item type
`igvf_portal_get_endpoint_params`	Discover available filter parameters for a collection
`igvf_portal_list_item_types`	List all valid item type names
`igvf_portal_facets`	Get aggregated facet counts for an item type

`igvf_portal_search`

Search the portal using free text and/or field filters.

igvf_portal_search(
    query: str = "",
    type: list[str] | None = None,
    limit: int | str = 25,
    sort: list[str] | None = None,
    field_filters: dict | None = None,
) -> str  # JSON

Parameter	Type	Description
`query`	str	Free-text search string
`type`	list[str]	Item types to filter by, e.g. `["SequenceFile"]`
`limit`	int or "all"	Max results (default 25)
`sort`	list[str]	Sort fields; prefix `-` for descending
`field_filters`	dict	Dotted field name → value filters. Values can be a single string or a list (repeated params). Use `!` suffix for negation, `gte:`/`lte:`/`gt:`/`lt:` for ranges.

Use real dotted field names (e.g. "file_set.@id"), not underscore param names. See igvf_portal_get_endpoint_params for the distinction.

# Single value
igvf_portal_search(type=["SequenceFile"], field_filters={"file_format": "fastq"})

# Multiple values (OR)
igvf_portal_search(type=["SequenceFile"], field_filters={"file_format": ["bam", "bed"]})

# Negation
igvf_portal_search(type=["SequenceFile"], field_filters={"file_format!": "fastq"})

# Range (lower bound)
igvf_portal_search(type=["SequenceFile"], field_filters={"file_size": "gte:1000000"})

# Range (bounded)
igvf_portal_search(type=["SequenceFile"], field_filters={"file_size": ["gte:1000000", "lte:5000000"]})

# Combined
igvf_portal_search(
    type=["SequenceFile"],
    field_filters={"file_format": "fastq", "file_set.@id": "/measurement-sets/IGVFDS1234ABCD/"},
    limit=5
)
# → {"total": 12340, "returned": 5, "results": [...]}

`igvf_portal_get_by_id`

Retrieve a single item by its @id, accession, or UUID.

igvf_portal_get_by_id(resource_id: str) -> str  # JSON

igvf_portal_get_by_id("IGVFFI1165AJSO")
igvf_portal_get_by_id("/sequence-files/IGVFFI1165AJSO/")

`igvf_portal_get_schema`

Return the JSON schema for an item type, describing all its fields.

igvf_portal_get_schema(item_type: str) -> str  # JSON schema

igvf_portal_get_schema("SequenceFile")

Use igvf_portal_list_item_types to get valid type names.

`igvf_portal_list_item_types`

Return a sorted list of all valid IGVF item type names (CamelCase).

igvf_portal_list_item_types() -> str  # JSON array

igvf_portal_list_item_types()
# → ["AccessKey", "AlignmentFile", "AnalysisSet", ...]

`igvf_portal_get_collection`

List items from a specific collection endpoint using underscore-style parameter names.

igvf_portal_get_collection(
    collection: str,
    query: str = "",
    limit: int | str = 25,
    sort: list[str] | None = None,
    field_filters: dict | None = None,
) -> str  # JSON

limit accepts an integer or "all" to return every result. field_filters uses underscore Python param names (e.g. "file_set_id"), not dotted field names. Use igvf_portal_get_endpoint_params to discover valid filter names.

igvf_portal_get_collection(
    "sequence_files",
    field_filters={"file_format": "fastq"},
    limit=10
)

`igvf_portal_get_endpoint_params`

Discover available filter parameters for a collection. Returns both underscore param names (for igvf_portal_get_collection) and dotted field names (for igvf_portal_search / igvf_portal_report).

igvf_portal_get_endpoint_params(endpoint: str) -> str  # JSON

igvf_portal_get_endpoint_params("sequence_files")
# → {
#     "endpoint": "sequence_files",
#     "filter_params": [
#       {"collection_param": "file_set_id", "search_field": "file_set.@id", "type": "str"},
#       ...
#     ]
#   }

igvf_portal_get_endpoint_params("?")  # list all available endpoints

`igvf_portal_download`

Download a single IGVF file to a local path.

igvf_portal_download(file_id: str, save_path: str) -> str  # JSON

igvf_portal_download("IGVFFI8092FZKL", "/tmp/IGVFFI8092FZKL.tsv")
# → {"saved_to": "/tmp/IGVFFI8092FZKL.tsv", "bytes": 204800}

`igvf_portal_batch_download`

Get download URLs for files belonging to matching FileSet items and save them to a local file. Only FileSet types are supported: MeasurementSet, AnalysisSet, AuxiliarySet, ConstructLibrarySet, CuratedSet, ModelSet, PredictionSet (and FileSet itself).

igvf_portal_batch_download(
    type: list[str],
    save_path: str,
    query: str = "",
    field_filters: dict | None = None,
) -> str  # JSON

igvf_portal_batch_download(
    type=["MeasurementSet"],
    save_path="/tmp/metadata.tsv",
    field_filters={"lab.@id": "/labs/ali-mortazavi/"}
)
# → {"saved_to": "/tmp/metadata.tsv", "bytes": 204800}

`igvf_portal_facets`

Get aggregated facet counts for an item type without fetching records. Useful for understanding data distribution before filtering.

igvf_portal_facets(
    type: list[str],
    query: str = "",
    field_filters: dict | None = None,
) -> str  # JSON

igvf_portal_facets(type=["SequenceFile"])
# → {"total": 45123, "facets": [{"field": "file_format", "title": "File Format", "terms": [...]}]}

`igvf_portal_report`

Generate a TSV-formatted report for an item type and save it locally.

igvf_portal_report(
    type: list[str],
    save_path: str,
    query: str = "",
    field_filters: dict | None = None,
) -> str  # JSON

igvf_portal_report(
    type=["SequenceFile"],
    save_path="/tmp/sequence_files.tsv",
    field_filters={"file_format": "fastq"}
)
# → {"saved_to": "/tmp/sequence_files.tsv", "bytes": 512000}

Agent Skill: `igvf-portal-facet-filter`

This repo includes an agent skill for guided faceted exploration. It is currently implemented as a Claude Code skill (.claude/skills/igvf-portal-facet-filter/SKILL.md) but the workflow it describes can be adapted for any coding agent.

Purpose

Guides the user through progressive filtering of an IGVF item type — starting from a high-level summary and narrowing down to a target set of records before fetching them.

Usage (Claude Code)

/igvf-portal-facet-filter SequenceFile

Or without an argument to be prompted for the item type.

Workflow

Calls igvf_portal_facets and shows the total item count plus a menu of available facet names — no values yet.
User picks facets to expand. The skill shows term values and counts only for those.
User picks a filter value. The skill re-queries with field_filters applied and shows the updated count and facet menu.
Steps 2–3 repeat until the user is ready to fetch records.
Fetches results with igvf_portal_search or igvf_portal_get_collection.

Example session

❯ /igvf-portal-facet-filter SequenceFile

● igvf-portal-mcp - igvf_portal_facets (MCP)(type: ["SequenceFile"])
  ⎿  {"result":"{\n  \"total\": 49945,\n  \"facets\": [...]\n  ...
     … +145 lines (ctrl+o to expand)

● Total SequenceFiles: 49,945

  Available facets (skipping single-value):
  Content Type, File Format, Illumina Read Type, Sequencing Kit,
  Sequencing Platform, Controlled Access, Data Use Limitation,
  File Set Type, Assay Term Names, Preferred Assay Titles, Taxa,
  Sample, Classification, Cellular Transformation Target, Disease,
  Modification, Treatment, Lab, Award

  Which facet(s) would you like to explore or filter on?

❯ file format

● From the data already fetched:

  fastq   47,981
  pod5     1,964

  Would you like to filter on one of these, explore another facet, or fetch results?

❯ sample

● From the data already fetched:

  putamen                            11,408
  posterior cingulate gyrus          11,292
  substantia nigra                   11,002
  middle temporal gyrus              10,768
  THP-1                               8,383
  cerebellar cortex                   8,348
  induced pluripotent stem cell       4,810
  Jurkat                              3,258
  peripheral blood mononuclear cell   3,213
  kidney                                758
  … +47 more

  Would you like to filter on a sample, explore another facet, or fetch results?

❯ Filter on fastq and kidney

● igvf-portal-mcp - igvf_portal_facets (MCP)(type: ["SequenceFile"],
                    field_filters: {"file_format":"fastq","file_set.samples.sample_terms.term_name":"kidney"})
  ⎿  {"result":"{\n  \"total\": 758, ...
     … +50 lines (ctrl+o to expand)

● 758 SequenceFiles match fastq + kidney. Mostly homogeneous:
  - Assay: single-nucleus RNA sequencing assay (Parse SPLiT-seq)
  - Taxa: Mus musculus / Lab: Ali Mortazavi, UCI

  Remaining multi-value facets:
  - Sequencing Kit:  NovaSeq 6000 S4 (480), NovaSeq X 25B (240), NextSeq P3 (28), NextSeq P4 (10)
  - Sequencing Platform: NovaSeq 6000 (480), NovaSeq X (240), NextSeq 2000 (38)
  - Classification: multiplexed sample (758), tissue/organ (758), primary cell (16)

  Would you like to filter further, or fetch the 758 results?

❯ filter on primary cell

● igvf-portal-mcp - igvf_portal_facets (MCP)(type: ["SequenceFile"],
                    field_filters: {"file_format":"fastq","file_set.samples.sample_terms.term_name":"kidney",
                                    "file_set.samples.classifications":"primary cell"})
  ⎿  {"result":"{\n  \"total\": 16, ...
     … +33 lines (ctrl+o to expand)

● 16 SequenceFiles match. Fully homogeneous — all single-value across every facet:
  - Format: fastq (reads) / Assay: Parse SPLiT-seq / Taxa: Mus musculus
  - Platform: Illumina NextSeq 2000 / Lab: Ali Mortazavi, UCI
  - Read types: R1 × 8, R2 × 8 / Controlled access: No

  Ready to fetch all 16 results?

❯ Download report and show me preview of first five rows

● igvf-portal-mcp - igvf_portal_report (MCP)(type: ["SequenceFile"],
                    save_path: "/tmp/sequence_files_kidney_primary.tsv",
                    field_filters: {"file_format":"fastq",
                                    "file_set.samples.sample_terms.term_name":"kidney",
                                    "file_set.samples.classifications":"primary cell"})
  ⎿  {"result":"{\n  \"saved_to\": \"/tmp/sequence_files_kidney_primary.tsv\",\n  \"bytes\": 3624\n}"}

● Report saved to /tmp/sequence_files_kidney_primary.tsv (3.6 KB). First 5 rows:

  ┌────────────────┬───────────┬────────────────┬──────────┐
  │   Accession    │ Read Type │    File Set    │  Status  │
  ├────────────────┼───────────┼────────────────┼──────────┤
  │ IGVFFI5210WGXF │ R1        │ IGVFDS1479KDWW │ released │
  ├────────────────┼───────────┼────────────────┼──────────┤
  │ IGVFFI7223DHDE │ R2        │ IGVFDS9176CSYH │ released │
  ├────────────────┼───────────┼────────────────┼──────────┤
  │ IGVFFI4385HOXT │ R2        │ IGVFDS1479KDWW │ released │
  ├────────────────┼───────────┼────────────────┼──────────┤
  │ IGVFFI2085FMMU │ R2        │ IGVFDS9176CSYH │ released │
  ├────────────────┼───────────┼────────────────┼──────────┤
  │ IGVFFI2033IHBM │ R1        │ IGVFDS9604WYAS │ released │
  └────────────────┴───────────┴────────────────┴──────────┘

  All 16 are fastq reads from Ali Mortazavi (UCI), released and validated,
  across 3 MeasurementSets (IGVFDS1479KDWW, IGVFDS9176CSYH, IGVFDS9604WYAS)
  with R1/R2 pairs for each.

This server cannot be installed

license - permissive license

quality - not tested

maintenance

How are these scores calculated?

Maintenance

–Maintainers

–Response time

–Release cycle

–Releases (12mo)

Commit activity

Resources

GitHub Repository

Need Help?

Related Servers

Unclaimed servers have limited discoverability.

Looking for Admin?

If you are the server author, to access and configure the admin panel.

Latest Blog Posts

Your AI Chatbot Just Exposed Your CEO's Salary to an Intern
By Om-Shree-0709 on July 2, 2026.
Agent Identity
MCP Security
OAuth Delegation
Why MCP Servers Need Execution Sandboxing (And Why Your Current Stack Isn't Enough)
By Om-Shree-0709 on June 30, 2026.
Agentic Ai
Prompt Injection
WebAssembly
Lightport: Open-Sourcing Glama's AI Gateway
By punkpeye on April 27, 2026.
OpenAI
open source

MCP directory API

We provide all the information about MCP servers via our MCP API.

curl -X GET 'https://glama.ai/api/mcp/v1/servers/IGVF-DACC/igvf-portal-mcp'

If you have feedback or need assistance with the MCP directory API, please join our Discord server