Tcia Remapping Skill

TCIA Remapping Skill

Overview

This Skill assists users in transforming their original clinical and imaging research data into the standardized TCIA (The Cancer Imaging Archive) data model. It leverages medical ontologies (NCIt, UBERON, SNOMED) and a structured conversational flow to ensure high-quality data mapping and standardization.

Target Data Model

The skill target consists of 12 potential TSV files defined in the resources/schema.json file:

• Program, Dataset, Subject, Procedure, File, Diagnosis, Investigator, Related_Work, Radiology, Histopathology, Multiplex_Imaging, Multiplex_Channels.

Conversational Workflow

To minimize user effort, the skill follows a tiered approach:

Phase 0: Dataset-Level Metadata Collection

Before remapping any source files, collect high-level metadata for the submission.

1. Sequential Interview: Collect information in the following order:

   • Start (Import): Offer the user the option to import an existing TCIA Proposal package (ZIP or TSV). If provided, use the import_proposal logic to pre-populate all subsequent fields.
   • Program: Focus on program_name and program_short_name (Required). Steering: Most users should be directed to use "Community" as their program unless they are part of a major NCI/NIH program (e.g., TCGA, CPTAC, APOLLO, Biobank).
   • Dataset: Focus on dataset_long_name and dataset_short_name (Required).
   • CICADAS: Guide the user through the CICADAS checklist to generate a high-quality dataset_abstract and dataset_description. (See tcia-cicadas-skill for detailed instructions).
   • Investigator: Collect details for one or more investigators. Ask for first_name, last_name, email, and organization_name (Required). Support multiple entries and ORCID lookups.
   • Related_Work: Collect DOI, title, publication_type, and relationship_type (Required). Support multiple entries and DOI lookups.

2. Handling Missing Info: If a required field is missing, prompt the user. If they don't have the information, acknowledge it and proceed to the next step.

3. Recap & Generation:

   • Provide a recap of all collected metadata for user review.
   • After approval, generate the corresponding TSV files (program.tsv, dataset.tsv, investigator.tsv, related_work.tsv) using the write_metadata_tsv function.

Phase 1: Structure Mapping & Organization

1. Initiation: Ask the user to upload their source data files.
2. Analysis: Identify existing columns and how they relate to the target entities.
3. Summary Recommendation: Provide a concise summary of the proposed mapping (e.g., "Source 'PtID' -> Target 'subject_id'").
4. Quick Approval: Allow the user to approve the entire structure mapping at once (e.g., "Everything looks good!") or flag specific items for discussion (e.g., "Discuss items 2 and 5").

Phase 2: Value Standardization (Permissible Values)

Once the structure is confirmed, address data content one TSV/column at a time:

1. Batch Validation: For each confirmed column, identify values that do not match the permissible lists in resources/permissible_values.json.
2. Ontology-Enhanced Matching: Use both fuzzy matching and knowledge of medical ontologies (NCIt, SNOMED, UBERON) to suggest corrections.
3. Focussed Discussion: Present a summary of auto-corrected values for quick approval, and list only the "hard" cases that need manual user intervention.
4. Handling "NOS": If a value is "NOS" (Not Otherwise Specified), suggest the most appropriate standard term from the ontology.

Key Rules

• Ontology Integration: When a direct match is missing, use NCIt or UBERON codes (found in permissible_values.json) to verify if a user's term is a synonym of a standard term.
• Tiered Interaction: Always provide a summary first. Don't ask about 100 values one by one; group them.
• Required Fields: Prioritize fields marked "R" in schema.json.
• Conflict Resolution: If an uploaded file contains data that contradicts previously provided metadata (e.g., a different Program Name), use check_metadata_conflict to alert the user. Ask them to confirm which value is correct.

Examples

Example: Tiered Approval

Claude: "I've mapped your 15 columns to the Subject and Diagnosis TSVs. Here is the summary: [Table]. Does this look good, or should we adjust specific mappings?" User: "Looks good, proceed to values." Claude: "Great. In 'primary_site', I've auto-matched 90% of your values. I have 3 values that need your attention: 'Lung, NOS', 'Chest wall', and 'Unknown'. Recommendations: [List]. How should we handle these?"

Resources

• resources/model/: NCI Imaging Submission Model files (MDF YAML format).
• resources/schema.json: Legacy property definitions (fallback).
• resources/permissible_values.json: Legacy valid values with ontology metadata (fallback).
• mdf_parser.py: Parser for the MDF model files.
• remap_helper.py: Utility script for fuzzy matching, data splitting, and conflict checking.
• orcid_helper.py: Utility for ORCID profile lookups and author parsing.

Converted and distributed by TomeVault — claim your Tome and manage your conversions.