Add D4D datasheet: CM4AI Cell Maps

[d4dassistant]

Summary

Created new D4D datasheet for Cell Maps for AI (CM4AI) based on comprehensive information from the CM4AI website and publications page.

Source Documentation

• Primary: https://cm4ai.org/
• Publications: https://cm4ai.org/publications/ (30+ papers identified)
• Data Releases: https://cm4ai.org/data-releases/
• Tools: https://cm4ai.org/tools/
• Training Resources: https://cm4ai.org/learning

Files Added

• data/sheets_d4dassistant/cm4ai_d4d.yaml - Comprehensive D4D YAML datasheet

Key Metadata Extracted

Dataset Overview

• Dataset ID: cm4ai-cell-maps
• Dataset Name: CM4AI Cell Maps for AI
• Description: Multimodal cellular maps integrating proteomics, imaging, and CRISPR perturbation data

Three Complementary Mapping Approaches

1. Proteomic Mass Spectrometry: Protein abundance, modifications, and interactions
2. Cellular Imaging: High-resolution microscopy of cellular morphology and localization
3. CRISPR/Cas9 Perturbation: Genetic knockouts and functional characterization

D4D Sections Populated

• ✅ Motivation: Purpose, tasks, funding (NIH Bridge2AI)
• ✅ Composition: Multimodal data types, sampling strategies, missing information
• ✅ Collection: Acquisition methods, collection mechanisms, data collectors, ethical reviews
• ✅ Preprocessing: Processing strategies for each modality, software tools used
• ✅ Uses: Recommended use cases, discouraged uses, impact considerations
• ✅ Distribution: Formats (mzML, TIFF, CSV, etc.), dates, licensing (open access principles)
• ✅ Maintenance: Maintainers, update frequency, versioning, retention policies

Ethical & Governance

• IRB approvals documented
• De-identification procedures described
• Consent processes included
• Open access licensing following NIH policies

Known Issues

Schema Validation

⚠️ Important Note: The current LinkML schema validation fails for this file AND for existing D4D files in the repository (including VOICE_d4d_alldocs.yaml, CM4AI_d4d_alldocs.yaml, etc.).

Validation issues observed:

• Existing D4D files use complex nested structures that don't match current schema expectations
• Schema appears to expect simple strings for most fields, but examples use structured objects
• This suggests schema and examples are out of sync

Recommendation: Schema review needed to align with the D4D file structure used across the repository.

How to Review

1. Read the YAML: Review data/sheets_d4dassistant/cm4ai_d4d.yaml for completeness and accuracy
2. Verify against sources: Compare metadata with CM4AI website and publications
3. Check required fields: Confirm id, name, title, and description are present and accurate
4. Assess coverage: Evaluate whether all major D4D sections are adequately populated

Metadata Quality

Strengths:

• Comprehensive coverage of all major D4D sections
• Based on official CM4AI project documentation
• Includes specific technical details (software tools, data formats, protocols)
• Documents ethical review and governance procedures
• References 30+ publications from CM4AI

Limitations:

• MCP tools (ARTL, WebFetch) encountered API issues, preventing direct paper analysis
• Some details inferred from project website rather than individual paper analysis
• Specific numeric details (participant counts, dataset sizes) not available from sources accessed
• Would benefit from review by CM4AI consortium members for technical accuracy

Next Steps

1. Review datasheet content for accuracy
2. Consider schema updates to support structured D4D metadata
3. Optionally: Enhance with specific metrics from publications once MCP tool access is resolved
4. Merge when satisfied with content

Related to: #91

---

🤖 Generated with D4D Assistant