# Sdrf:review > Use when the user wants a comprehensive quality review of an SDRF file, a PR review of an SDRF submission, or a quality score assessment. - **Type:** Skill - **Install:** `agentstack add skill-bigbio-sdrf-skills-sdrf-review` - **Verified:** Yes — security-reviewed for prompt injection and unsafe behavior - **Seller:** [bigbio](https://agentstack.voostack.com/s/bigbio) - **Installs:** 0 - **Category:** [Agent Skills](https://agentstack.voostack.com/c/agent-skills) - **Latest version:** 0.1.0 - **License:** MIT - **Upstream author:** [bigbio](https://github.com/bigbio) - **Source:** https://github.com/bigbio/sdrf-skills/tree/main/skills/sdrf-review - **Website:** https://sdrf.quantms.org ## Install ```sh agentstack add skill-bigbio-sdrf-skills-sdrf-review ``` Requires the [AgentStack CLI](https://agentstack.voostack.com/docs/cli). Works with Claude Code, Cursor, and any MCP-compatible agent. ## About # SDRF Review Workflow You are performing a comprehensive quality review of an SDRF file — like a peer reviewer would for a PRIDE submission or a community annotation PR. ## Step 1: Load Context 1. **Read the SDRF** content 2. **Detect templates** from metadata or content 3. **If PXD available** → fetch project context + publication: ```text mcp PRIDE → get_project_details(project_accession="PXD######") Extract PMID → mcp PubMed → get_article_metadata([pmid]) ``` 4. **If GitHub PR** → read the diff to understand what changed ## Step 2: Run Full Validation (sdrf-validate workflow) Apply the complete validation checklist from the sdrf-validate skill. Read `spec/sdrf-proteomics/TERMS.tsv` for column definitions and `spec/sdrf-proteomics/sdrf-templates/templates.yaml` for template metadata. Collect all errors and warnings. ## Step 3: Cross-Reference with Publication If a publication is available: - Does the sample count in SDRF match the paper? - Are all conditions from the paper represented? - Do the instruments match? - Are demographics (age, sex) consistent with the paper? - Is `characteristics[developmental stage]` supported by the cohort description even if age is reported only at group level? - Are tissue types correctly annotated? When Europe PMC full text is available, do not inspect raw XML directly. First run: `python scripts/europepmc_fulltext.py PMC_ID --section methods --section results --section discussion --format text` or use `--format json` when structured links, captions, or accession detection will help the review. Flag any discrepancies: ```text DISCREPANCY: Paper says "24 patients" but SDRF has 20 unique source names. DISCREPANCY: Paper mentions "hippocampus and temporal cortex" but SDRF only has "brain". ``` For technical metadata (instrument, tolerances, modifications, DDA/DIA), consider recommending `/sdrf:techrefine` — techsdrf can verify these parameters directly from the raw MS files, which is more reliable than cross-referencing with the publication. ### Conflict Resolution When SDRF and paper/PRIDE disagree: - **Sample count mismatch**: Check if some samples were excluded (QC failure, outliers). Paper may report enrolled patients while SDRF has analyzable samples. Check supplementary tables. - **Instrument mismatch**: PRIDE might say "Q Exactive" while paper says "Q Exactive HF". The paper is usually more specific — update SDRF to match the paper's instrument model. - **Tissue specificity**: If paper says "hippocampus" but SDRF says "brain", update SDRF to the more specific term from the paper. - **Demographic mismatch**: If paper has a demographics table, prioritize it. SDRF might have been filled from incomplete metadata. - **Cohort-only demographics**: If the paper reports only cohort summaries, `developmental stage` may still be supportable, but do not force per-sample `age`, `sex`, or `ethnicity` without an individual-level mapping table. - **File count mismatch**: Some files in PRIDE may be non-raw (search results, FASTA, etc.). Compare only raw files. ## Step 4: Cross-Reference with PRIDE If a PXD accession is available: - Do file names in SDRF match files in PRIDE? ```text mcp PRIDE → get_project_files(project_accession="PXD######") ``` REST fallback: ```text GET https://www.ebi.ac.uk/pride/ws/archive/v3/projects/PXD######/files/all ``` - Does the organism match? - Does the instrument match? - Are all raw files accounted for? If PRIDE exposes no raw files and the dataset is hosted by MassIVE, use the deterministic helper: ```bash python -m tools massive-files PXD016117 --mode raw --format tsv ``` Treat this as a fallback for reconstructing defensible `comment[data file]` values when the repository metadata is incomplete. ## Step 5: Design Analysis Analyze the experimental design: - Are factor values properly defined? - Is the comparison clear? - Are replicates adequate? - Are there potential batch effects? (instrument × condition confounding) - Are there confounders? (age × disease, sex × treatment) ## Step 6: Quality Score Calculate and present an overall quality score: ```text # SDRF Quality Review: PXD012345 # Date: [date] # Templates: ms-proteomics, human, clinical-metadata ## Quality Score: 78/100 ### Breakdown: Completeness: 85/100 ████████░░ Specificity: 70/100 ███████░░░ Consistency: 90/100 █████████░ Standards: 65/100 ██████░░░░ Design Clarity: 80/100 ████████░░ ### Errors Found: 3 1. [ERROR] Row 12: UNIMOD:21 used for Acetyl (should be UNIMOD:1) 2. [ERROR] Missing required column: characteristics[biological replicate] 3. [ERROR] "cancer" is too generic — specify cancer subtype ### Warnings: 5 1. [WARN] Case inconsistency in characteristics[sex]: "Male" vs "male" 2. [WARN] Not using latest template version (v1.0.0 → v1.1.0 available) 3. [WARN] Missing recommended column: characteristics[cell type] 4. [WARN] Age format "58 years" should be "58Y" 5. [WARN] Missing SDRF metadata columns (comment[sdrf version]) ### Improvements Suggested: 3 1. [IMPROVE] Disease term could be more specific (EFO:0000311 "cancer" → EFO:0000305 "breast carcinoma") 2. [IMPROVE] Add characteristics[developmental stage] — 80% of human studies include it 3. [IMPROVE] Consider adding comment[sdrf annotation tool] for provenance ### Cross-Reference: Publication match: ✓ Sample count matches paper PRIDE file match: ✓ All 240 raw files present Instrument match: ✓ Q Exactive HF confirmed ### Verdict: NEEDS MINOR FIXES Fix the 3 errors and this SDRF is ready for submission. ``` ## Step 7: Actionable Next Steps Provide clear next steps: 1. List fixes in priority order 2. Offer to auto-fix what can be auto-fixed (`/sdrf:fix`) 3. Identify what needs human input (e.g., missing demographics) 4. Suggest running final validation after fixes 5. If the SDRF is for a ProteomeXchange dataset and the verdict is VALID or NEEDS MINOR FIXES: suggest contributing the annotation via `/sdrf:contribute {PXD}` to the `sdrf-annotated-datasets` community repository ## Source & license This open-source skill is cataloged on AgentStack and links to its original source — we do not rehost the code. - **Author:** [bigbio](https://github.com/bigbio) - **Source:** [bigbio/sdrf-skills](https://github.com/bigbio/sdrf-skills) - **License:** MIT - **Homepage:** https://sdrf.quantms.org Install and usage instructions live in the source repository linked above. ## Pricing - **Free** — Free ## Security capabilities Automated source analysis of v0.1.0 — what this tool can access: - **Network access:** no - **Filesystem access:** no - **Shell / process execution:** no - **Environment & secrets:** no - **Dynamic code execution:** no *"Yes" means the capability is present in the source — more access means more to trust, not that it is unsafe.* ## Versions - **0.1.0** — security scan: passed — Imported from the upstream source. ## Links - Listing page: https://agentstack.voostack.com/l/skill-bigbio-sdrf-skills-sdrf-review - Seller: https://agentstack.voostack.com/s/bigbio - Browse the marketplace: https://agentstack.voostack.com/browse --- Listed on AgentStack — the marketplace for AI agent skills and MCP servers. Every listing is security-reviewed. Creators keep 70%.