Add cpg0044-uhler-multicells and onboard-dataset skill

.claude/skills/onboard-dataset/SKILL.md

@@ -0,0 +1,222 @@
+---
+name: onboard-dataset
+description: Helps Cell Painting Gallery maintainers onboard new dataset contributions. Handles Phase 1 setup including assigning cpg identifiers, updating prefixes.md, and creating GitHub Discussion tracking issues. Use when a contributor contacts you about adding a new dataset or when you need to initiate the dataset contribution workflow.
+allowed-tools:
+  - Read
+  - Write
+  - Edit
+  - Bash
+  - Glob
+  - Grep
+  - AskUserQuestion
+---
+
+# Dataset Onboarding Skill (Phase 1)
+
+You are helping a Cell Painting Gallery maintainer onboard a new dataset contribution. This skill handles **Phase 1 only**: initial setup, identifier assignment, and tracking issue creation.
+
+## Context
+
+The Cell Painting Gallery follows a staging-to-production workflow. Contributors contact maintainers (@erinweisbart or @shntnu) to initiate contributions. Your job is to:
+
+1. Gather contributor information
+2. Assign a unique cpg#### identifier
+3. Update the prefixes registry
+4. Create a GitHub Discussion for tracking
+5. Provide next steps
+
+## Step 1: Gather Contributor Information
+
+**First, check if the user has provided context** (email thread, message, etc.) in their initial request.
+
+### If context is provided
+
+- **Parse the text** to extract:
+  - Assay type (standard Cell Painting or variation)
+  - Approximate data size
+  - Components to contribute (images, analysis, backend, profiles, metadata, etc.)
+  - Institution/affiliation
+  - Suggested project names/tags
+  - Contributor name/contact
+  - Whether this is a new dataset or addition to existing dataset
+
+- **Present what you found** with confidence levels, like:
+
+  ```
+  📧 Extracted from provided context:
+  ✓ Assay type: [what you found]
+  ✓ Data size: [what you found]
+  ✓ Institution: [what you found]
+  ? Project tag: [if ambiguous or unclear]
+  ? Components: [if not explicitly listed]
+  ```
+
+- **Ask interactive questions ONLY for**:
+  - Missing information
+  - Ambiguous details that need clarification
+  - Confirmation of assumptions
+
+### If no context is provided
+
+Use the `AskUserQuestion` tool to gather all required information interactively.
+
+### Required Information
+
+1. **Assay type**:
+   - Standard Cell Painting
+   - Variation (describe it)
+   - Must label multiple cellular compartments/organelles
+
+2. **Approximate data size**: Total size estimate (e.g., "5 TB", "500 GB")
+
+3. **Components to contribute**:
+   - Major (at least one required): `images`, `analysis`, `backend`, `load_data_csv`, `profiles`
+   - Required: `metadata`
+   - Optional: `pipelines`, `qc`, `profiles_assembled`, `segmentation`
+
+4. **Institution identifier**:
+   - Examples: `broad`, `anonymous`, `source_4`
+   - Must be lowercase, can use underscores
+
+5. **Suggested project tag**:
+   - Typically 1-2 words describing the project
+   - Examples: `jump`, `cmqtl`, `resistance`, `caie-drugresponse`
+   - Sometimes includes author last name
+   - Must be lowercase, use hyphens to separate words
+
+6. **Contributor contact**: Name and/or GitHub handle
+
+7. **Dataset type**: New dataset or addition to existing dataset?
+   - If existing, which cpg#### prefix does it extend?
+
+## Step 2: Assign cpg#### Identifier
+
+1. **Read** `documentation/prefixes.md` to see all existing identifiers
+2. **Find the next available cpg number**:
+   - Parse the list (format: `cpg0000-tag`, `cpg0001-tag`, etc.)
+   - Find highest number
+   - Add 1 to get next available
+3. **Format the full identifier**: `cpg####-suggested-tag`
+   - Validate naming convention:
+     - Lowercase letters only
+     - Numbers allowed
+     - Hyphens allowed (no underscores in tag portion)
+     - No spaces or special characters
+     - Max 64 characters total
+4. **Show the maintainer** the proposed identifier and ask for confirmation
+
+## Step 3: Update prefixes.md
+
+1. **Add the new identifier** to `documentation/prefixes.md`
+2. **Maintain numerical order** in the table
+3. **Use the Edit tool** to add the line in the correct position
+4. **Show the maintainer** what you added
+
+## Step 4: Create GitHub Discussion
+
+1. **Read the template** from `.claude/skills/onboard-dataset/templates/discussion_template.md`
+   - This is based on the official template: https://github.com/broadinstitute/cellpainting-gallery/discussions/66
+   - Enhanced with structured Dataset Information section for easier tracking
+
+2. **Customize the template** with the information gathered:
+   - Fill in all Dataset Information fields (identifier, contributor, assay type, size, components, type)
+   - Fill in institution identifier
+   - Fill in champion name (the maintainer using this skill, unless specified otherwise)
+   - Check the first workflow item (prefixes.md updated)
+
+3. **Create the discussion** using the helper script:
+   - Use simple title format: `cpg####-tag` or `cpg####-tag (brief description)`
+     - Examples from existing discussions:
+       - `cpg0020-varchamp`
+       - `cpg0029-chroma-pilot`
+       - `cpg0016-JUMP (deep-learning generated profiles via Mesmer)`
+   - First, write the customized template to a temporary file (e.g., `discussion_body.md`)
+   - Then run the helper script:
+
+   ```bash
+   .claude/skills/onboard-dataset/scripts/create_discussion.sh \
+     discussion_body.md \
+     "cpg####-tag"
+   ```
+
+   The script will:
+   - Properly escape the body content for GraphQL
+   - Use the correct repository ID (`R_kgDOGULepQ`) and category ID (`DIC_kwDOGULepc4CZL49` for "Dataset descriptions")
+   - Create the discussion via GitHub GraphQL API
+   - Return the discussion URL and number
+
+   **Optional**: If you need to verify category IDs or use a different category:
+
+   ```bash
+   gh api graphql -f query='
+   query {
+     repository(owner: "broadinstitute", name: "cellpainting-gallery") {
+       id
+       discussionCategories(first: 10) {
+         nodes {
+           id
+           name
+         }
+       }
+     }
+   }'
+   ```
+
+   Then pass the custom IDs to the script:
+
+   ```bash
+   .claude/skills/onboard-dataset/scripts/create_discussion.sh \
+     discussion_body.md \
+     "Title" \
+     "REPO_ID" \
+     "CATEGORY_ID"
+   ```
+
+   **Result**: The script will display the discussion URL and number when successful
+
+## Step 5: Generate Next Steps Summary
+
+Provide the maintainer with a brief confirmation:
+
+### ✅ Dataset onboarding initiated
+
+- **Assigned**: cpg####-tag
+- **Updated**: `documentation/prefixes.md`
+- **Tracking**: [Discussion URL]
+
+→ All dataset details and next steps are captured in the GitHub Discussion
+
+## Important Notes
+
+- **DO NOT** proceed to Phase 2 (credentials/upload) or Phase 3 (validation/transfer) - those are handled separately
+- **DO NOT** commit changes to git without showing the maintainer first
+- **DO NOT** push to remote - the maintainer will review and push when ready
+- If the contributor is adding to an **existing dataset**, note that in the discussion and skip updating README.md (will be updated when complete)
+- Always validate that the cpg number is unique before proceeding
+- The tag portion of the identifier should be meaningful and memorable
+
+## Error Handling
+
+- If `gh` CLI is not available, the script will fail with a clear error message. Provide the template body and title for manual creation via GitHub web UI
+- If `jq` is not installed, the script will fail with a clear error. Install it with `brew install jq` (macOS) or appropriate package manager
+- If the script fails to create the discussion, it will display the GraphQL response for debugging
+- If prefixes.md cannot be read, ask the maintainer to verify file location
+- If the suggested tag violates naming conventions, suggest alternatives
+- If unclear about any information, always ask rather than guess
+
+## Files and Scripts You'll Use
+
+- `documentation/prefixes.md` - Registry of all cpg identifiers
+- `.claude/skills/onboard-dataset/templates/discussion_template.md` - GitHub Discussion template (based on [#66](https://github.com/broadinstitute/cellpainting-gallery/discussions/66))
+- `.claude/skills/onboard-dataset/scripts/create_discussion.sh` - Helper script for creating GitHub discussions
+- `.git/` - For git operations (checking branch, status)
+
+## Commands You May Use
+
+- `.claude/skills/onboard-dataset/scripts/create_discussion.sh` - Create GitHub discussion with proper escaping
+- `gh api graphql` - Query/mutate GitHub data via GraphQL API (optional, script handles discussion creation)
+- `git status` - Check repository state
+- `git add` - Stage changes (only with maintainer approval)
+- `gh repo view` - Verify repository details
+
+Remember: This is Phase 1 only. Focus on gathering information, assigning identifiers, and setting up tracking. The maintainer will handle credentials, upload coordination, and validation in subsequent phases.

.claude/skills/onboard-dataset/scripts/create_discussion.sh

@@ -0,0 +1,84 @@
+#!/bin/bash
+# create_discussion.sh - Create a GitHub Discussion for dataset onboarding
+#
+# Usage: create_discussion.sh <body_file> <title> [repo_id] [category_id]
+#
+# Arguments:
+#   body_file    - Path to markdown file containing the discussion body
+#   title        - Discussion title (e.g., "cpg####-tag" or "cpg####-tag (description)")
+#   repo_id      - GitHub repository ID (default: R_kgDOGULepQ for broadinstitute/cellpainting-gallery)
+#   category_id  - Discussion category ID (default: DIC_kwDOGULepc4CZL49 for "Dataset descriptions")
+
+set -e  # Exit on error
+
+# Default values
+DEFAULT_REPO_ID="R_kgDOGULepQ"
+DEFAULT_CATEGORY_ID="DIC_kwDOGULepc4CZL49"
+
+# Parse arguments
+BODY_FILE="${1}"
+TITLE="${2}"
+REPO_ID="${3:-$DEFAULT_REPO_ID}"
+CATEGORY_ID="${4:-$DEFAULT_CATEGORY_ID}"
+
+# Validate inputs
+if [[ -z "$BODY_FILE" ]] || [[ -z "$TITLE" ]]; then
+    echo "Error: Missing required arguments" >&2
+    echo "" >&2
+    echo "Usage: $0 <body_file> <title> [repo_id] [category_id]" >&2
+    echo "" >&2
+    echo "Example:" >&2
+    echo "  $0 discussion_body.md 'cpg0023-test'" >&2
+    exit 1
+fi
+
+if [[ ! -f "$BODY_FILE" ]]; then
+    echo "Error: Body file not found: $BODY_FILE" >&2
+    exit 1
+fi
+
+# Check for required commands
+if ! command -v gh &> /dev/null; then
+    echo "Error: gh CLI is not installed or not in PATH" >&2
+    exit 1
+fi
+
+if ! command -v jq &> /dev/null; then
+    echo "Error: jq is not installed or not in PATH" >&2
+    exit 1
+fi
+
+# Prepare the body with proper JSON escaping
+BODY=$(jq -Rs . < "$BODY_FILE")
+
+# Create the GraphQL mutation
+RESPONSE=$(gh api graphql -f query="
+mutation {
+  createDiscussion(input: {
+    repositoryId: \"$REPO_ID\"
+    categoryId: \"$CATEGORY_ID\"
+    title: $(echo "$TITLE" | jq -R .)
+    body: $BODY
+  }) {
+    discussion {
+      url
+      id
+      number
+    }
+  }
+}")
+
+# Extract and display the discussion URL
+DISCUSSION_URL=$(echo "$RESPONSE" | jq -r '.data.createDiscussion.discussion.url')
+DISCUSSION_NUMBER=$(echo "$RESPONSE" | jq -r '.data.createDiscussion.discussion.number')
+
+if [[ "$DISCUSSION_URL" != "null" ]] && [[ -n "$DISCUSSION_URL" ]]; then
+    echo "✅ Discussion created successfully!"
+    echo "   URL: $DISCUSSION_URL"
+    echo "   Number: #$DISCUSSION_NUMBER"
+    exit 0
+else
+    echo "❌ Failed to create discussion" >&2
+    echo "Response: $RESPONSE" >&2
+    exit 1
+fi

.claude/skills/onboard-dataset/templates/discussion_template.md

@@ -0,0 +1,58 @@
+<!-- Dataset onboarding discussion template
+     This is the authoritative template for dataset contribution discussions.
+     Previously maintained at: https://github.com/broadinstitute/cellpainting-gallery/discussions/66
+
+     Enhanced with structured Dataset Information section for maintainer workflow.
+-->
+
+## Dataset Information
+
+- **Identifier**: cpg####-tag
+- **Contributor**: [name/contact]
+- **Assay Type**: [Standard Cell Painting / variation description]
+- **Approximate Size**: [size estimate]
+- **Components**: [images, analysis, backend, profiles, metadata, etc.]
+- **Type**: [New dataset / Addition to cpg####]
+
+Profile/data repo = [Link to profile/data repo or TBD]
+Publication/analysis repo = [Link to publication/analysis repo or TBD]
+Imaging Platform project tag = [YYYY_MM_DD_ProjectName] DELETE IF EXTERNAL CONTRIBUTOR
+Institution identifier = [institution identifier]
+Imaging Platform internal "champion" = [Maintainer name or specify if external]
+
+**Workflow**
+
+- [ ] Maintainer adds dataset identifier to [prefixes](https://github.com/broadinstitute/cellpainting-gallery/blob/main/documentation/prefixes.md) document.
+- [ ] Champion or Contributor fills out the metadata collection form (<https://airtable.com/shrVxz9DcoMlDoCBI>)
+- [ ] Champion checks that metadata completely filled out in Project Profiler Database (Imaging Platform internal use only)
+- [ ] Champion asks Erin/Shantanu to  1) open the prefix for transfer to `staging-cellpainting-gallery` and 2) create credentials that allow upload. Each credential is for a single prefix on the staging bucket; multiple people can assume that same credential.
+
+**Transfer data to `staging-cellpainting-gallery`**
+Instructions for using the staging bucket credentials are [here](https://broadinstitute.github.io/cellpainting-gallery/uploading_to_cpg.html).
+For each data subtype, Champion marks off as complete or explain why it is not being included:
+
+- [ ] Image acquisition complete and uploaded (`/images`)
+- [ ] Feature extraction complete and uploaded (`/workspace/analysis`)
+- [ ] Backend complete and uploaded (`/workspace/backend`)
+- [ ] Load_data complete and uploaded (`/workspace/load_data_csv`)
+- [ ] Profiling complete and uploaded (`/workspace/profiles`)
+- [ ] Metadata complete and uploaded (`/workspace/metadata`)
+
+**Project completion**
+These are the final steps to make a dataset fully public:
+
+- [ ] Maintainer or champion Run [validation script](http://github.com/broadinstitute/cpg/cpgdata) to ensure data is completely uploaded (When script is out of alpha)
+- [ ] Maintainer or champion updates cellpainting-gallery/README.md to add dataset to README.
+- [ ] Maintainer transfers from `staging-cellpainting-gallery` to `cellpainting-gallery`.
+- [ ] Erin/Shantanu disable staging prefix, delete data from staging, delete upload credentials.
+
+Optional preparation for publication:
+
+- [ ] Run Distributed-BioFormats2Raw to create .ome.zarr files
+- [ ] Upload (meta)data to IDR (images remain hosted in cellpainting-gallery)
+- [ ] Make IDR entry public
+
+Once a paper is published:
+
+- [ ] Maintainer or champion updates [Publications](https://github.com/broadinstitute/cellpainting-gallery/docs/publications.md) (In PR by data generator)
+- [ ] Contributor adds the paper to <https://github.com/cytodata/awesome-cytodata>