/deepspot-m

name: deepspot-m
description: Transcriptome-wide virtual spatial transcriptomics from H&E histology with DeepSpot-M. Scores a 224x224 tile and returns per-gene log1p-CPM values for any HGNC symbols you ask for, with a CSV, a report and a reproducibility bundle.
license: MIT
metadata:
  version: "0.3.0"
  # The weights carry their own terms. Nested under `metadata` because the
  # agentskills spec allows no other top-level key than the six it names, and
  # a top-level `model_license` fails `agentskills validate`.
  model_license: cc-by-nc-sa-4.0
  author: Kalin Nonchev
  domain: spatial-transcriptomics
  tags:
    - spatial-transcriptomics
    - histology
    - gene-expression
    - foundation-model
    - digital-pathology
    - h-and-e
  inputs:
    - name: input_file
      type: file
      format:
        - png
        - jpg
        - jpeg
        - tif
        - tiff
      description: One 224x224 H&E tile cut at native (~20x) resolution
      required: true
  outputs:
    - name: report
      type: file
      format:
        - md
      description: Per-gene expression report with the upstream limitations attached
    - name: result
      type: file
      format:
        - json
      description: Machine-readable per-gene log1p-CPM values and run parameters
    - name: tables
      type: file
      format:
        - csv
      description: Gene table, one row per gene
    - name: reproducibility
      type: directory
      format:
        - dir
      description: commands.sh, environment.yml and checksums.sha256
  dependencies:
    python: ">=3.11"
    packages:
      - deepspotm>=1.0,<2
      - Pillow>=9.0
  demo_data:
    - path: examples/demo_tile.png
      description: Synthetic 224x224 H&E-like tile
    - path: examples/demo_expression.json
      description: Offline fixture standing in for a gene panel readout
  endpoints:
    cli: python skills/deepspot-m/deepspot_m.py --input {input_file} --output {output_dir}
  openclaw:
    requires:
      bins:
        - python3
    always: false
    emoji: "🧬"
    homepage: https://github.com/ratschlab/DeepSpotM
    os:
      - darwin
      - linux
    install:
      - kind: pip
        package: deepspotm
    trigger_keywords:
      - virtual spatial transcriptomics
      - gene expression from histology
      - spatial transcriptomics from H&E
      - predict gene expression from a tissue image
      - DeepSpot-M

🧬 DeepSpot-M Virtual Spatial Transcriptomics

You are **deepspot-m**, a specialised ClawBio agent that turns an H&E histology tile into virtual spatial transcriptomics. You score one 224x224 tile with the DeepSpot-M foundation model and report per-gene log1p-CPM values for the gene symbols the user names.

Trigger

**Fire this skill when the user says any of:**

• "virtual spatial transcriptomics"
• "predict gene expression from histology"
• "spatial transcriptomics from H&E"
• "what genes are expressed in this tissue image"
• "score this tile for BRAF and COL1A1"
• "run DeepSpot-M on this tile"
• "gene expression map from a slide"
• "H&E to transcriptome"

**Do NOT fire when:**

• The user wants cells counted or outlined in an image. That is `cell-detection`.
• The user already has a measured spot-count table and wants region labels. That is `marker-dominance-mapper`.
• The user wants differential expression between conditions from a count matrix. That is `rnaseq-de`.
• The user wants single-cell clustering or embedding of an AnnData object. That is `scrna-orchestrator` or `scrna-embedding`.
• The user asks for TCGA bulk expression lookups. That is `xena-tcga-gene-query`.

Why This Exists

• **Without it**: Reading expression off an archived slide means running a spatial assay on the tissue, which most samples never get.
• **With it**: One archived H&E tile yields per-gene values in one command, entirely on the local machine.
• **Why ClawBio**: The call goes to a published model with released weights, pinned to one checkpoint, and every run leaves a reproducibility bundle behind.

This is a research tool, not a substitute for measurement. The model card publishes no per-gene accuracy figure, and neither does the preprint abstract, so this skill quotes none. Read the preprint for the evaluation before treating any number here as a finding, and see `## Safety` for the limitations upstream states.

Core Capabilities

1. **Score a tile**: Map one 224x224 H&E tile to per-gene log1p-CPM values. 2. **Query genes**: Ask for any HGNC symbols in the released panel and get only those, which is faster than scoring the whole transcriptome. 3. **Choose an embedding source**: Route gene queries through Evo 2, Orthrus, ProtT5, scGPT or Apertus embeddings. 4. **Check the tile**: Flag tiles that are near-white background or essentially colourless before reporting numbers for them. 5. **Report**: Write `report.md`, `result.json`, a gene CSV and a reproducibility bundle.

Scope

One skill, one task. This skill scores a single H&E tile and writes gene values. It does not read whole-slide images, tile them, register sections, call cells, or compute spatial statistics. For a whole slide, tile it first and call this skill per tile, or use `examples/predict_wsi.py` from the upstream repository.

Input Formats

| Format | Extension | Required Properties | Example | |--------|-----------|---------------------|---------| | PNG | `.png` | Exactly 224x224 px, H&E stained | `examples/demo_tile.png` | | JPEG | `.jpg`, `.jpeg` | Exactly 224x224 px, H&E stained | `tile.jpg` | | TIFF | `.tif`, `.tiff` | Exactly 224x224 px, H&E stained | `tile.tif` |

Tiles must be exactly 224x224 pixels. The skill checks the dimensions and stops with an explicit message when they differ. Upstream cuts tiles on a 224-pixel grid at native (~20x) resolution (source: upstream README, `### Command line`).

**On microns per pixel**: no microns-per-pixel or magnification figure appears on the model card, and the only magnification upstream states anywhere is the