Tencent scPROTEIN Single-Cell Proteome Analysis

Problem

Single-cell proteomics data are often distorted by batch effects, missing values, and uncertainty. Running a public app directly also requires handling CLI authentication, project/environment/storage-bucket configuration, Nextflow version selection, app-name conflicts, and inputs that are not visible to the job container. This skill narrows the scPROTEIN workflow into a controlled path for denoising, uncertainty estimation, embedding generation, and cell-type representation.

How It Works

• Locked collection: only scPROTEIN Collection (Nextflow) is supported, with hardcoded subapps scPROTEIN_stage1 and scPROTEIN_stage2; the collection AppId cannot be used directly.
• Preflight checks: runs whoami, config show, and trial-quota checks; C-end users are blocked when days or daily runs are exhausted.
• Import safety: checks for duplicate app names before omics run, prompts renaming on conflict, and requires --nf-version for Nextflow subapps.
• File inputs: local files must be explicitly approved and uploaded to COS before being used as parameters.
• Failure handling: interprets DUPLICATE_APP_NAME, PARAM_MERGE_FAILED, and MISSING_NF_VERSION, and may reuse an orphan app via --app without deleting existing apps.

Boundaries

This is not a general omics CLI skill. It refuses requests for apps outside this collection and points to omics-task-skill. It is intended for engineers with a Tencent Health Omics Platform account, configured project/environment/COS bucket, and a need to reproduce the two-stage scPROTEIN pipeline.