メニュー
Prepare a spatial transcriptomics method for contribution to an OpenProblems benchmark (e.g. task_ist_preprocessing). Use when the user wants to submit/contribute a method, open a PR to a task_* repo, or check submission readiness.
Codex または Claude でインストール この Prompt をコピーして Codex、Claude、または他のアシスタントに貼り付けると、Skill ページを確認してインストールできます。
SOC 職業分類に基づく
| name | openproblems-submission |
| description | Prepare a spatial transcriptomics method for contribution to an OpenProblems benchmark (e.g. task_ist_preprocessing). Use when the user wants to submit/contribute a method, open a PR to a task_* repo, or check submission readiness. |
Goal: a Viash component that conforms to a task's API, builds cleanly, passes the test profile, and is reproducible — ready for a PR.
Repo setup. Clone the target task_* repo and initialize the common/
git submodule (git submodule update --init --recursive). Read the repo's
CONTRIBUTING / README for current conventions.
Conform to the component API. Your config.vsh.yaml must __merge__ the
correct comp_<stage>.yaml interface and match its inputs/outputs and
info/metadata fields. Run analyze_workflow_configuration on the config.
Data contract. Inputs/outputs are SpatialData (zarr) / AnnData as the stage
requires. Validate test data with validate_spatial_data and confirm the
elements the stage needs are present (check_spatial_data_compatibility for
multi-file). State raw-vs-normalized expectations explicitly.
Reproducibility. Build on the shared base images
(openproblems/base_python:1 / base_r:1) with __merge__-ed setup
partials rather than raw python:*/r-base images. Pin base-image tags and
package versions — no unpinned latest. Provide the standard metadata
(label, summary, description, links, references/doi) and document
every parameter with a default and biological rationale.
Build & test locally (server doesn't run these — use the terminal):
viash ns buildviash run config.vsh.yaml -- --input <test>.zarr --output tmp/out.zarrviash test config.vsh.yaml — a unit test (test_resources) is expected
for every componentnextflow run main.nf -profile test,dockerDependency sanity. Run analyze_workflow_dependencies across the workflow
files to catch missing/conflicting containers or libraries.
PR hygiene. Include test data + expected output, document parameter choices, and confirm CI (GitHub Actions) builds and runs the component.
build_openproblems_method, run_openproblems_benchmark, and
validate_openproblems_submission are roadmap. Use the validation/analysis
tools that exist, and run the build/benchmark steps via local CLIs.Diagnose a failing Nextflow pipeline run in an OpenProblems benchmark. Use when a `nextflow run` fails, a process errors out, or the user shares a .nextflow.log / .command.err and asks why it broke.
Validate and inspect spatial transcriptomics data files (SpatialData/zarr/AnnData) before any analysis. Use when the user points at a spatial data file, reports a loading/format error, or asks whether data is suitable for a method.
Author or modify a Viash component (config.vsh.yaml + script) for an OpenProblems method, especially a task_ist_preprocessing stage. Use when the user wants to add/implement a spatial method, scaffold a component, or fix a config.vsh.yaml.