菜单
Load when estimating cell-type proportions in bulk RNA-seq samples from a single-cell or signature-matrix reference. Skip if the data is already single-cell (no deconvolution needed) or for spatial deconvolution (use spatial-deconv).
默认使用会先检查来源的 Prompt；你也可以切换为直接命令，或下载本地副本。
决定是否安装前，请先阅读 SKILL.md，以及 SkillsMP 当前展示的配套文件。
用 Codex 或 Claude 帮你安装 复制这段 Prompt，粘贴到 Codex、Claude 或其他助手里，让它检查 Skill 页面并帮你完成安装。
直接命令不会经过审查 Prompt；运行前请先检查来源。
npx skills add https://github.com/mdbabumiamssm/LLMs-Universal-Life-Science-and-Clinical-Skills- --skill bulkrna-deconvolution命令会保持在同一行。复制前请横向滚动并检查完整内容。
想先保存到本地？可下载 SkillsMP 当前能够提供的文件。
Operate MedSAM2 for promptable segmentation of 3D medical images and medical videos, including CT lesion propagation, MRI volumes, RECIST-guided prompts, efficient CPU-oriented variants, training, and 3D Slicer integration. Use when generating or validating volumetric masks from sparse prompts or propagating masks through image slices or video frames.
Build reproducible healthcare imaging pipelines with Project MONAI for DICOM, NIfTI, pathology, and multidimensional imaging tasks including preprocessing, augmentation, training, sliding-window inference, evaluation, model bundles, labeling, and deployment. Use when implementing medical image classification, segmentation, registration, detection, generative, or foundation-model workflows in PyTorch.
Operate Google TxGemma prediction and chat models for therapeutic property prediction across small molecules, proteins, nucleic acids, diseases, targets, and cell lines. Use when formatting Therapeutics Data Commons tasks, choosing TxGemma model size or variant, running local or Model Garden inference, fine-tuning on private therapeutic data, or evaluating TxGemma in drug-discovery workflows.
基于 SOC 职业分类
正在显示 SKILL.md
| name | bulkrna-deconvolution |
| description | Load when estimating cell-type proportions in bulk RNA-seq samples from a single-cell or signature-matrix reference. Skip if the data is already single-cell (no deconvolution needed) or for spatial deconvolution (use spatial-deconv). |
| version | 0.3.0 |
| author | OmicsClaw |
| license | MIT |
| tags | ["bulkrna","deconvolution","NNLS","cell-type-proportion"] |
Run on a bulk RNA-seq cohort when you have a reference (single-cell profile or signature matrix) and want per-sample cell-type proportions. Built-in NNLS solver is the only backend currently implemented; the wrapper does not call CIBERSORTx or MuSiC.
| Input | Format | Required |
|---|---|---|
| Bulk count matrix | .csv (gene × sample) | yes (or --demo) |
| Reference | --reference CSV (signature matrix) or .h5ad (sc reference) | yes (or --demo) |
| Output | Path | Notes |
|---|---|---|
| Proportions | tables/proportions.csv | sample × cell_type, rows sum to 1 |
| Dominant types | tables/dominant_types.csv | per-sample top cell type |
| Stacked-bar plot | figures/proportions_stacked.png | per-sample composition |
| Heatmap | figures/proportions_heatmap.png | sample × cell_type intensity |
| Report | report.md + result.json | always |
bulkrna_deconvolution.py:368 raises ValueError if --input missing without --demo).:370 raises ValueError if --reference missing without --demo; :76 raises FileNotFoundError if path doesn't exist).:131 raises ValueError if no overlap).scipy.optimize.nnls per sample to estimate per-cell-type weights, then row-normalise to proportions.tables/proportions.csv + tables/dominant_types.csv.--reference is REQUIRED for non-demo runs. Unlike most bulkrna skills, this one needs two inputs. bulkrna_deconvolution.py:370 raises with "--reference is required when not using --demo" if you forget; no silent fallback.:131 raises ValueError when bulk and reference share zero gene IDs (typical cause: bulk uses Ensembl, reference uses HGNC symbols). Pre-run bulkrna-geneid-mapping to harmonise.--method flag. Despite the skill catalog historically advertising CIBERSORTx and MuSiC bridges, the script (bulkrna_deconvolution.py:347-358 argparser) accepts only --input, --output, --demo, --reference. The summary dict (:163-172) records n_genes_shared, n_samples, n_cell_types, cell_types, proportions_df, dominant_types, mean_proportions, residuals — no method field, because there is no choice.residuals (saved in result.json["residuals"]) measure how well the linear combination explains the bulk profile. Sanity-check that residuals are small relative to library size; large residuals indicate the reference is missing a major cell type from the bulk.python omicsclaw.py run bulkrna-deconvolution --demo
python omicsclaw.py run bulkrna-deconvolution \
--input counts.csv --reference signature.csv --output results/
python omicsclaw.py run bulkrna-deconvolution \
--input counts.csv --reference scref.h5ad --output results/
references/parameters.md — every CLI flag and tuning hintreferences/methodology.md — NNLS solver, gene-overlap requirement, residual interpretationreferences/output_contract.md — exact output directory layoutbulkrna-geneid-mapping (run upstream to harmonise gene IDs), bulkrna-trajblend (parallel: per-sample pseudotime placement using the same NNLS proportions plus nearest-neighbour mapping), spatial-deconv (spatial-side sibling: spot-level proportions with multiple methods)