=== YAML Fixing Applied ===
id: "doi:10.18130/V3/F3TD5R"
name: Cell Maps for Artificial Intelligence - June 2025 Data Release (Beta)
title: Cell Maps for Artificial Intelligence - June 2025 Data Release (Beta)
description: This dataset is the June 2025 Data Release of Cell Maps for Artificial Intelligence (CM4AI; CM4AI.org), the Functional Genomics Grand Challenge in the NIH Bridge2AI program. This Beta release includes perturb-seq data in undifferentiated KOLF2.1J iPSCs; SEC-MS data in undifferentiated KOLF2.1J iPSCs, iPSC-derived NPCs, neurons, cardiomyocytes, and treated and untreated MDA-MB-468 breast cancer cells; and IF images in MDA-MB-468 breast cancer cells in the presence and absence of chemotherapy (vorinostat and paclitaxel). Some datasets are under temporary pre-publication embargo; interrogated protein sets across modalities incompletely overlap; computed cell maps are not included in this release. Long-term preservation is planned in the University of Virginia Dataverse.
language: en
page: "https://dataverse.lib.virginia.edu/dataset.xhtml?persistentId=doi:10.18130/V3/F3TD5R"
doi: "doi:10.18130/V3/F3TD5R"
issued: 2025-07-01
created_on: 2025-02-27
version: "2.0"
license: CC BY-NC-SA 4.0
keywords:
  - AI
  - artificial intelligence
  - machine learning
  - Bridge2AI
  - CM4AI
  - induced pluripotent stem cell
  - iPSC
  - KOLF2.1J
  - neural progenitor cell
  - NPC
  - neuron
  - cardiomyocyte
  - breast cancer
  - MDA-MB-468
  - perturb-seq
  - perturbation sequencing
  - single-cell RNA sequencing
  - scRNAseq
  - CRISPR/Cas9
  - protein-protein interaction
  - SEC-MS
  - size exclusion chromatography
  - mass spectroscopy
  - affinity purification
  - AP-MS
  - protein localization
  - subcellular imaging
  - vorinostat
  - paclitaxel
created_by:
  - CM4AI Consortium
  - University of Virginia Dataverse
purposes:
  - response: Create AI-ready multimodal datasets to support research in functional genomics within the NIH Bridge2AI CM4AI project.
tasks:
  - response: AI model training on multimodal cellular data (scRNA-seq/perturb-seq, SEC-MS proteomics, IF imaging).
  - response: Cellular process and architecture analysis across disease-relevant cell lines and treatment conditions.
  - response: Study of cell architectural changes and interactions under chemotherapy or genetic perturbations.
addressing_gaps:
  - response: Provide standardized, de-identified, AI-ready cell maps inputs across multiple modalities for disease-relevant human cell lines; predicted cell maps will be included in future releases.
creators:
  - principal_investigator:
      name: Trey Ideker
      affiliation:
        name: University of California San Diego
    affiliation:
      name: University of California San Diego
  - principal_investigator:
      name: Emma Lundberg
      affiliation:
        name: Stanford University
    affiliation:
      name: Stanford University
  - principal_investigator:
      name: Nevan Krogan
      affiliation:
        name: University of California San Francisco
    affiliation:
      name: University of California San Francisco
  - principal_investigator:
      name: Prashant Mali
      affiliation:
        name: University of California San Diego
    affiliation:
      name: University of California San Diego
  - principal_investigator:
      name: Andrej Sali
      affiliation:
        name: University of California San Diego
    affiliation:
      name: University of California San Diego
  - principal_investigator:
      name: Tim Clark
      affiliation:
        name: University of Virginia
    affiliation:
      name: University of Virginia
funders:
  - grantor:
      name: National Institutes of Health
    grant:
      name: Bridge2AI Functional Genomics Grand Challenge
      grant_number: 1OT2OD032742-01
instances:
  - representation: Confocal microscopy immunofluorescence images (ICC-IF) of breast cancer cell line MDA-MB-468 under untreated and chemotherapy conditions (vorinostat, paclitaxel); 464 proteins of interest.
    instance_type: Multichannel images per field-of-view with protein-of-interest channel plus cellular markers.
    data_type: Image data packaged as ZIP archives (multichannel IF image files; provenance graphs in HTML).
    label: No supervised labels provided; images stratified by protein target and treatment condition.
    missing_information:
      - missing:
          - Predicted/computed cell maps are not included in this release.
          - Exact per-protein image counts are not enumerated in the landing page.
        why_missing:
          - Computed maps planned for future releases; summary counts not provided in the page content.
    sampling_strategies:
      - is_sample:
          - Yes
        is_random:
          - No stated random sampling; selection based on proteins of interest.
        source_data:
          - De-identified human cancer cell line (MDA-MB-468) grown under specified treatments.
        is_representative:
          - No (does not represent all biological variants in the population).
        why_not_representative:
          - Derived from commercially available de-identified cell lines and targeted protein panels.
  - representation: Size exclusion chromatography mass spectrometry (SEC-MS) proteomics characterizing protein complexes/interactions.
    instance_type: Proteins/protein complexes profiled across multiple human cell types and conditions.
    data_type: Mass spectrometry data (SEC-MS) with modality-specific RO-Crate metadata; external deposition in MassIVE.
    label: No explicit labels; interaction/complex inference downstream.
    sampling_strategies:
      - is_sample:
          - Yes
        source_data:
          - De-identified human iPSCs, iPSC-derived NPCs, neurons, cardiomyocytes, and MDA-MB-468 cancer cells (treated/untreated).
        is_representative:
          - No (targeted protein sets; cell line-based models).
        why_not_representative:
          - Interrogated proteins incompletely overlap across modalities and do not cover all biological variants.
  - representation: Perturb-seq (CRISPRi) single-cell RNA-seq in undifferentiated KOLF2.1J iPSCs.
    instance_type: Single cells profiled by scRNA-seq following genetic perturbations.
    data_type: scRNA-seq reads and processed matrices; external deposition via SRA BioProject.
    label: Perturbation identities and conditions; gene expression profiles per cell.
    sampling_strategies:
      - is_sample:
          - Yes
        source_data:
          - KOLF2.1J iPSCs under CRISPRi perturbations.
        is_representative:
          - No (cell line and perturbation panel constrained).
relationships:
  - description:
      - Multimodal alignment possible across shared targets and conditions (protein panels and treatments overlap across IF and SEC-MS; perturb-seq targets overlap incompletely).
splits:
  - description:
      - No recommended train/dev/test splits provided; users should create modality-appropriate splits.
anomalies:
  - description:
      - None reported; users should consider modality-specific technical noise (e.g., MS batch effects, imaging artifacts).
external_resources:
  - external_resources:
      - Sequence Read Archive (SRA) Data: NCBI BioProject (perturb-seq)
      - Mass Spectrometry Data (Human iPSCs): MassIVE Repository
      - Mass Spectrometry Data (Human Cancer Cells): MassIVE Repository
    future_guarantees:
      - External repositories (NCBI SRA, MassIVE) provide archival services, but persistence and identifiers should be verified from those resources.
    archival:
      - Long-term preservation of this release in University of Virginia Dataverse.
    restrictions:
      - External repository terms and licenses may apply to specific sub-deposits.
confidential_elements:
  - description:
      - No patient or clinical record data; de-identified human cell lines only.
content_warnings:
  - warnings:
      - None noted.
subpopulations:
  - identification:
      - Human cell types/lines and conditions: KOLF2.1J iPSCs, iPSC-derived NPCs, neurons, cardiomyocytes, MDA-MB-468 cells; untreated and treated (vorinostat, paclitaxel).
    distribution:
      - Not quantified in the landing page; modality coverage and protein panels vary.
sensitive_elements:
  - description:
      - Biological (omic and image) data derived from de-identified human cell lines; no direct personal identifiers.
is_deidentified:
  description:
    - Human Subjects: No; De-identified Samples: Yes.
    - FDA Regulated: No.
acquisition_methods:
  - description:
      - Directly observed measurements: confocal microscopy imaging (ICC-IF), SEC-MS proteomics, scRNA-seq for perturb-seq.
      - No subject-reported data.
      - No model-inferred primary labels in this release (computed maps not included).
    was_directly_observed: Yes
    was_reported_by_subjects: No
    was_inferred_derived: Limited; downstream analyses may infer interactions/complexes but primary data are observational.
    was_validated_verified: Not specified on the landing page; provenance graphs provided for imaging workflows.
collection_mechanisms:
  - description:
      - Hardware/software and procedures: ICC-IF staining and confocal microscopy (Lundberg Lab, Stanford); size exclusion chromatography mass spectrometry (SEC-MS); CRISPRi perturbation with single-cell RNA-seq (perturb-seq). RO-Crate metadata and provenance graphs accompany subsets.
data_collectors:
  - description:
      - CM4AI consortium teams including Stanford University (Lundberg Lab), University of California San Diego, University of California San Francisco, University of Virginia, and collaborators listed as dataset authors.
collection_timeframes:
  - description:
      - Data creation date: 2025-02-27; initial release publication date: 2025-07-01; additional IF image ZIP archives published 2025-10-22.
ethical_reviews:
  - description:
      - Data Governance Committee: Jillian Parker (jillianparker@health.ucsd.edu). Ethical Review: Vardit Ravitsky (ravitskyv@thehastingscenter.org) and Jean-Christophe Bélisle-Pipon (jean-christophe_belisle-pipon@sfu.ca).
data_protection_impacts:
  - description:
      - Data are de-identified cell line-based; no direct data subjects. Potential downstream inferences should be evaluated by users.
preprocessing_strategies:
  - description:
      - Modality-specific pipelines (details referenced via RO-Crate metadata and provenance graphs); computed cell maps are not included in this release.
cleaning_strategies:
  - description:
      - Not detailed on landing page; modality-specific QA/QC expected per RO-Crate and provenance documentation.
labeling_strategies:
  - description:
      - Imaging channels include DAPI (nuclei), calreticulin (ER), tubulin (microtubules), and protein-of-interest antibody; conditions labeled by treatment. No supervised labels provided for ML tasks in this release.
raw_sources:
  - description:
      - External raw/processed data available via SRA (perturb-seq) and MassIVE (SEC-MS); imaging archives provided as ZIPs with provenance.
existing_uses:
  - description:
      - Related publications: 2024 preprints describing CM4AI data and analyses (e.g., bioRxiv doi:10.1101/2024.05.21.589311; doi:10.1101/2024.11.03.621734).
other_tasks:
  - description:
      - Benchmarking multimodal integration methods; representation learning across imaging and omics; condition and perturbation effect prediction.
future_use_impacts:
  - description:
      - Cell line-based data and targeted protein panels limit generalizability; incomplete overlap of protein sets across modalities may affect integration strategies.
discouraged_uses:
  - description:
      - Prohibited: Use in clinical decision-making or any patient care context without appropriate regulatory oversight and approval.
distribution_formats:
  - description:
      - HTML datasheet (release-ro-crate-datasheet.html)
      - RO-Crate metadata (JSON)
      - ZIP archives (IF images per condition)
      - HTML provenance graphs (per modality/condition)
distribution_dates:
  - description:
      - 2025-07-01 (dataset publication)
      - 2025-10-22 (IF image ZIP archives)
license_and_use_terms:
  description:
    - License: CC BY-NC-SA 4.0. Community norms expect proper credit via citation as shown on the dataset page.
maintainers:
  - description:
      - Long-term preservation hosted by University of Virginia Dataverse (institutional funds committed). Point of contact noted on dataset page; Data Governance Committee contact: Jillian Parker (CM4AI).
errata:
  - description:
      - This is a Beta/interim release; computed cell maps not yet included; some datasets under temporary pre-publication embargo.
updates:
  description:
    - Dataset will be regularly updated and augmented through November 2026; updates planned on a quarterly basis. Long-term preservation in the University of Virginia Dataverse.
version_access:
  description:
    - Older versions retained by Dataverse; republishing and versioning managed via the UVA Dataverse platform.
is_tabular: "no"
subsets:
  - name: release-ro-crate-datasheet.html
    title: Release Datasheet (HTML)
    description: HTML datasheet summarizing key release information.
    md5: 599c9ece9b88b3ce797b82463b4a1eb4
    media_type: text/html
    format: JSON
    path: release-ro-crate-datasheet.html
  - name: release-ro-crate-metadata.json
    title: Release RO-Crate Metadata (JSON)
    description: RO-Crate metadata with pointers to sub RO-Crates.
    md5: 99f9e00053bff3020fd9832a3a518bbb
    media_type: application/json
    format: JSON
    path: release-ro-crate-metadata.json
  - name: cm4ai-ifimages-mda-mb-468-paclitaxel.zip
    title: IF Images — MDA-MB-468 treated with paclitaxel (ZIP)
    description: Spatial localization of 464 proteins in MDA-MB-468 cells treated with paclitaxel; ICC-IF staining and confocal microscopy (DAPI, calreticulin, tubulin, protein-of-interest channels).
    md5: 0d972b80744344ddeede516a0cf6e3d7
    media_type: application/zip
    path: Images/cm4ai-ifimages-mda-mb-468-paclitaxel.zip
    is_data_split: no
    is_subpopulation: Treatment-specific subset
  - name: cm4ai-ifimages-mda-mb-468-untreated.zip
    title: IF Images — MDA-MB-468 untreated (ZIP)
    description: Spatial localization of 464 proteins in MDA-MB-468 cells (untreated); ICC-IF staining and confocal microscopy.
    md5: a98affcc05429650c6bb3906cd836d55
    media_type: application/zip
    path: Images/cm4ai-ifimages-mda-mb-468-untreated.zip
    is_data_split: no
    is_subpopulation: Condition-specific subset
  - name: cm4ai-ifimages-mda-mb-468-vorinostat.zip
    title: IF Images — MDA-MB-468 treated with vorinostat (ZIP)
    description: Spatial localization of 464 proteins in MDA-MB-468 cells treated with vorinostat; ICC-IF staining and confocal microscopy.
    md5: ad4e68ccc14b0f3349dad3321e7b81b2
    media_type: application/zip
    path: Images/cm4ai-ifimages-mda-mb-468-vorinostat.zip
    is_data_split: no
    is_subpopulation: Treatment-specific subset
  - name: Images-paclitaxel-provenance-graph.html
    title: IF Images — paclitaxel provenance graph (HTML)
    description: Provenance graph for paclitaxel imaging workflow; download to view properly.
    md5: e38e63e4c8dfc5808a5ffa2d7829fc38
    media_type: text/html
    path: Images/paclitaxel/Images-paclitaxel-provenance-graph.html
  - name: Images-untreated-provenance-graph.html
    title: IF Images — untreated provenance graph (HTML)
    description: Provenance graph for untreated imaging workflow; download to view properly.
    md5: 1a3b510f74d3f8647e07c6559ce64ee8
    media_type: text/html
    path: Images/untreated/Images-untreated-provenance-graph.html
  - name: Images-vorinostat-provenance-graph.html
    title: IF Images — vorinostat provenance graph (HTML)
    description: Provenance graph for vorinostat imaging workflow; download to view properly.
    md5: 58935fe4e254b31d33fed019f24c7668
    media_type: text/html
    path: Images/vorinostat/Images-vorinostat-provenance-graph.html
  - name: mass-spec-cancer-cells-provenance-graph.html
    title: Mass spectrometry — cancer cells provenance graph (HTML)
    description: Provenance graph for SEC-MS workflow in human cancer cells; download to view properly.
    md5: 931ad9b552562024cb84ebe62d1f1838
    media_type: text/html
    path: mass-spec/cancer-cells/mass-spec-cancer-cells-provenance-graph.html
  - name: mass-spec-cancer-cells-ro-crate-metadata.json
    title: Mass spectrometry — cancer cells RO-Crate metadata (JSON)
    description: RO-Crate metadata for cancer cell SEC-MS datasets.
    md5: 3a7063bb391ea5e05a32ba5da5f4b2f8
    media_type: application/json
    format: JSON
    path: mass-spec/cancer-cells/mass-spec-cancer-cells-ro-crate-metadata.json
use_repository:
  - description:
      - Dataset usage metrics available via DataCite/Make Data Count on the Dataverse landing page.