菜单
Load when stratifying patients by gene expression and testing for survival differences (Kaplan-Meier + Cox) in bulk RNA-seq. Skip if no time-to-event clinical data exists, or for non-bulk cohorts (single-cell / spatial survival is not supported).
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。
基于 SOC 职业分类
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.
Evaluate and operate released Profluent OpenCRISPR gene-editing systems, especially OpenCRISPR-1, for controlled research workflows using its published Cas9-like protein, compatible guide RNA designs, protocols, licensing, specificity testing, and experimental validation. Use when comparing OpenCRISPR-1 with SpCas9, planning nonclinical editing studies, or assessing use in nuclease, nickase, deactivated, base, prime, or epigenome-editing contexts.
Operate CZI TranscriptFormer cross-species generative single-cell models to produce cell embeddings, contextual gene embeddings, likelihoods, zero-shot classifiers, disease-state representations, and regulatory analyses from raw-count AnnData files. Use when selecting TF-Sapiens, TF-Exemplar, or TF-Metazoa, processing in- or out-of-distribution species, or scaling embedding extraction across GPUs.
Operate ByteDance Protenix-v2 for open biomolecular structure prediction of proteins, antibodies, nucleic acids, ligands, and complexes using JSON inputs, MSA and template features, constraints, and inference-time sampling. Use when running Protenix locally or through its server, comparing AlphaFold3-style open models, or building reproducible co-folding evaluations.
| name | bulkrna-survival |
| description | Load when stratifying patients by gene expression and testing for survival differences (Kaplan-Meier + Cox) in bulk RNA-seq. Skip if no time-to-event clinical data exists, or for non-bulk cohorts (single-cell / spatial survival is not supported). |
| version | 0.3.0 |
| author | OmicsClaw |
| license | MIT |
| tags | ["bulkrna","survival","Kaplan-Meier","Cox","hazard-ratio","clinical"] |
Run on a bulk RNA-seq cohort with paired clinical survival data (time-to-event + censoring) when you want to ask "does high vs low expression of gene X predict survival?". Default workflow: per-gene median-cutoff stratification, log-rank p-value, Kaplan-Meier curve, and Cox proportional-hazards hazard ratio.
| Input | Format | Required |
|---|---|---|
| Expression matrix | .csv (gene × sample) | yes (or --demo) |
| Clinical data | .csv (sample, time, event cols) via --clinical | yes (or --demo) |
| Genes to test | --genes TP53,BRCA1 (comma-separated) | optional, defaults to all in expression matrix |
| Output | Path | Notes |
|---|---|---|
| Per-gene survival stats | tables/survival_results.csv | columns: gene, cutoff, n_high, n_low, hazard_ratio, log_rank_chi2, log_rank_pval, median_survival_high, median_survival_low |
| KM curves | figures/km_<gene>.png | one per gene tested |
| Forest plot | figures/forest_plot.png | HR + CI across genes |
| Report | report.md + result.json | summary contains n_genes and per-gene results array (bulkrna_survival.py:153-160) |
--genes (or all):
bulkrna_survival.py:630 if gene not in expression matrix.--cutoff-method (default median; alt optimal finds the maxstat cut).:326 ("Heavy censoring (X%). KM tail estimates may be unreliable.") when the censoring rate exceeds 80%.survival package first; fall back to Python lifelines (:626 warns "R survival not available (...); using Python fallback.").tables/survival_results.csv.bulkrna_survival.py:630 logs a warning per missing gene and continues. After the run, count the rows in tables/survival_results.csv (or inspect result.json["results"]) and compare against the --genes list — a typo'd or wrong-namespace gene produces no obvious error.--cutoff-method optimal p-values are NOT corrected for multiple testing. The optimal cutoff scans all possible cuts and picks the maximally separating one, which inflates Type I error. Reported log-rank p-values are raw — apply Bonferroni / BH correction externally if you scan many genes.(events_high / time_high) / (events_low / time_low) (bulkrna_survival.py:328-333), not a Cox proportional-hazards regression coefficient. This estimator is biased when proportional-hazards holds with unequal exposure — for publication-grade HRs, re-fit a proper Cox model in R or lifelines against the same stratification.:626 warns and falls back to a NumPy log-rank implementation when R survival isn't importable; the per-gene HR estimator is the same simple events/time ratio in both cases, but the chosen backend isn't recorded in the summary dict — only in the warning log. Verify R availability before relying on the result for downstream papers.:326 fires when ≥80% of patients are censored; the printed median survival numbers are dominated by extrapolation past the last event time. Treat median_survival_* as "≥ X" rather than a point estimate when the corresponding gene's censoring rate is high.python omicsclaw.py run bulkrna-survival --demo
python omicsclaw.py run bulkrna-survival \
--input expression.csv --clinical clinical.csv \
--genes TP53,BRCA1,EGFR --output results/
python omicsclaw.py run bulkrna-survival \
--input expression.csv --clinical clinical.csv \
--genes TP53 --cutoff-method optimal --output results/
references/parameters.md — every CLI flag and tuning hintreferences/methodology.md — KM + log-rank + Cox theory, R vs Python backend differences, optimal-cutoff caveatsreferences/output_contract.md — exact output directory layoutbulkrna-de (parallel: differential expression — survival adds the time-to-event dimension), bulkrna-coexpression (parallel: module-level survival via eigengene if traits include time-to-event)