Add D4D datasheet: Cell Maps for Artificial Intelligence (CM4AI)

[d4dassistant]

Summary

Created new D4D datasheet for Cell Maps for Artificial Intelligence (CM4AI) based on comprehensive documentation from:

• https://cm4ai.org (main project website)
• https://cm4ai.org/publications/ (38 related publications)
• https://dataverse.lib.virginia.edu/dataverse/CM4AI (data repository)
• https://cm4ai.org/data-releases/ (quarterly data releases)
• https://cm4ai.org/standards/ (FAIRSCAPE metadata standards)
• https://cm4ai.org/product-documentation/ (data products documentation)
• https://dataverse.lib.virginia.edu/citation?persistentId=doi:10.18130/V3/B35XWX (March 2025 Beta release)

Files Added

• data/extracted_by_column/CM4AI/cm4ai_comprehensive_d4d.yaml - D4D YAML datasheet
• src/html/output/D4D_-_CM4AI_Dataverse_v3_human_readable.html - HTML preview

Validation

• ✅ Schema validation passed
• ✅ Required fields populated (id, name)
• ✅ YAML syntax valid
• ✅ HTML preview generated

Key Metadata Extracted

• Dataset ID: cm4ai-cell-maps
• Dataset Name: Cell Maps for Artificial Intelligence
• Purpose: Map the spatiotemporal architecture of human cells for interpretable genotype-phenotype learning using multimodal approaches (proteomics, imaging, CRISPR perturbation)
• Composition: 53,788 immunofluorescent images, 1,374 protein interactions, 1,792 proteins investigated, 11,739 genes targeted
• Data Volume: 22.7 TB total
• Distribution: University of Virginia Dataverse with quarterly releases (CC BY-NC-SA 4.0)
• AI-Ready: RO-Crate format with FAIRSCAPE metadata framework and Evidence Graph Ontology provenance
• Funding: NIH Bridge2AI grant 1OT2OD032742-01
• Institutions: UCSD, UCSF, Stanford, UVA, Yale, UT Austin, UAB, Simon Fraser University, Hastings Center

Dataset Features

• Multimodal data: Affinity purification mass spectrometry (AP-MS), immunofluorescence imaging, CRISPR/Cas9 perturbation screens, SEC-MS
• Cell types: MDA-MB-468 breast cancer cells, iPSCs, iPSC-derived NPCs, neurons, cardiomyocytes
• Treatment conditions: Untreated, paclitaxel-treated, vorinostat-treated
• Target proteins: 100 chromatin modifiers + 100 metabolic enzymes (cancer, neuropsychiatric, cardiac disorders)
• Standards: FAIR principles, Schema.org vocabulary, JSON-LD serialization, W3C PROV provenance

How to Review

1. View HTML preview: Open src/html/output/D4D_-_CM4AI_Dataverse_v3_human_readable.html in a browser for human-readable format
2. Check YAML: Review data/extracted_by_column/CM4AI/cm4ai_comprehensive_d4d.yaml for completeness and accuracy
3. Validate sources: Compare against original documentation URLs listed above
4. Verify required fields: Ensure id and name are present and accurate
5. Check optional fields: Confirm populated fields are accurate; additional fields can be added based on schema

Notes

• This is a minimal valid datasheet focusing on core required fields and key descriptive metadata
• Additional D4D schema fields (purposes, tasks, instances, subsets, etc.) can be added in future updates as the schema structure supports more complex nested objects
• All information extracted directly from official CM4AI sources
• Dataset uses cell lines only (no human subjects data)
• Quarterly data releases ongoing through project completion

Publications Referenced

Key publications that describe the dataset:

• Clark et al. (2024) "Cell Maps for Artificial Intelligence: AI-Ready Maps of Human Cell Architecture from Disease-Relevant Cell Lines" bioRxiv doi:10.1101/2024.05.21.589311
• Nourreddine et al. (2024) "A Perturbation Cell Atlas of Human Induced Pluripotent Stem Cells" bioRxiv
• 36 additional publications listed at https://cm4ai.org/publications/

Related to: #66

---

🤖 Generated with D4D Assistant