组学分析技能指数

其他 社区

技能解读

当前原文的自动摘译,尚未人工审校。

核心单细胞技能

适用于最常见单细胞分析任务的高优先级、可操作的工作流程。技能索引:single_cell/SKILL.md 技能:质量控制:过滤、双峰检测、归一化、QC 指标细胞类型注释:基于标记和基于参考的标签分配

基因组选择

用于在 scRNA-seq 和空间转录组学 (HVG/DE/RF/scGeneFit/SpaPROS) 中设计基因 panel 的端到端工作流程,包括子 panel 发现、共识评分、生物学完成和基准测试。技能文件夹:gene_panel_selection/ 使用时机:

空间组学

空间转录组学绘图、插补和 3D 可视化技能。技能索引:spatial/SKILL.md 技能:单细胞到空间映射:使用 MOSCOT 将 scRNA-seq 映射到空间数据,以进行基因插补和细胞类型转移

单细胞基础模型 (SCFM)

用于与基础模型(scGPT、Geneformer、UCE、scBERT 等)嵌入/集成的工作流程和模型参考。技能索引:scfm/SKILL.md 何时使用:您需要 FM 嵌入(例如,obsm["X_uce"]、obsm["X_scGPT"])

数据库访问

用于以编程方式查询基因组数据库、下载测序数据和访问大规模单细胞数据集的工具。技能索引:database_access/SKILL.md 涵盖的工具:gget:23个模块,用于查询Ensembl、NCBI、UniProt、COSMIC、OpenTargets等。

流狐档案 运行与权限由流狐静态检查推断,不代表作者声明或实际测试结果。
领域
通用
作者声明的智能体
未找到明确声明;不据此推断已兼容或已测试
静态检查
88 / 100 · 启发式扫描,不代表运行安全
运行环境证据
未执行;当前没有隔离运行测试证据
作者 / 版本 / 许可证
@aristoteleo · 未声明许可证
上下文用量评级
低消耗
接入复杂程度
即装即用
是否需要外部接口密钥
不需要
兼容的系统
未声明(默认跨平台)
运行环境要求
未检出明确要求
检测到的文件与系统行为
  • 只读
网络行为
仅限本地
安装命令数
无(仅作为资料)

需要注意: 未声明工具权限范围;实际权限由运行环境决定。

作者说明文件(原文) SKILL.md

# Agent Skills for Omics Data Analysis

Best practices and workflows for single-cell and spatial omics analysis.
Load the relevant skill files when performing specific analysis tasks.

## Core Single-Cell Skills

High-priority, actionable workflows for the most common single-cell analysis tasks.

**Skill index**: [single_cell/SKILL.md](./single_cell/SKILL.md)

**Skills**:
- **Quality Control**: Filtering, doublet detection, normalization, QC metrics
- **Cell Type Annotation**: Marker-based and reference-based label assignment
- **Trajectory Inference**: Pseudotime, lineage tracing, RNA velocity

---

## Gene Panel Selection

End-to-end workflow for designing gene panels in scRNA-seq and spatial
transcriptomics (HVG/DE/RF/scGeneFit/SpaPROS), with sub-panel discovery,
consensus scoring, biological completion, and benchmarking.

**Skill folder**: [gene_panel_selection/](./gene_panel_selection/)

**When to use**:
- Designing a gene panel for spatial transcriptomics
- Benchmarking existing panels (ARI/NMI/Silhouette + UMAP)
- **IMPORTANT**: When doing gene panel selection, **strictly** follow this workflow

---

## Spatial Omics

Skills for spatial transcriptomics mapping, imputation, and 3D visualization.

**Skill index**: [spatial/SKILL.md](./spatial/SKILL.md)

**Skills**:
- **Single-Cell to Spatial Mapping**: Map scRNA-seq to spatial data with MOSCOT
  for gene imputation and cell type transfer
- **3D Spatial Visualization**: Interactive 3D plots and rotating animations
  with PyVista

**When to use**:
- You have paired scRNA-seq and spatial transcriptomics data
- You want to impute genes or transfer cell type labels to spatial coordinates
- Your spatial data has 3D coordinates and you want to visualize them

---

## Single-Cell Foundation Models (SCFM)

Workflow and model reference for embedding/integration with foundation models
(scGPT, Geneformer, UCE, scBERT, etc.).

**Skill index**: [scfm/SKILL.md](./scfm/SKILL.md)

**When to use**:
- You want FM embeddings (e.g., `obsm["X_uce"]`, `obsm["X_scGPT"]`)
- You need model selection based on gene ID scheme and species
- You want a validation-first workflow before heavy inference

---

## Database Access

Tools for querying genomic databases, downloading sequencing data, and
accessing large-scale single-cell datasets programmatically.

**Skill index**: [database_access/SKILL.md](./database_access/SKILL.md)

**Tools covered**:
- **gget**: 23 modules for querying Ensembl, NCBI, UniProt, COSMIC, OpenTargets, etc.
- **iSeq**: CLI for downloading from GSA, SRA, ENA, DDBJ, GEO
- **CZ CELLxGENE Census**: API for 217M+ single-cell observations

---

## Upstream Processing

Technology-specific pipelines for processing raw sequencing data into
analysis-ready count matrices.

**Skill index**: [upstream_processing/SKILL.md](./upstream_processing/SKILL.md)

**Technologies covered**:
- **nf-core Pipelines**: 143+ Nextflow pipelines for scRNA-seq, spatial, bulk,
  ATAC-seq, ChIP-seq, variant calling
- **OpenST**: Open-source spatial transcriptomics processing pipeline

---

## General Data Analysis

Cross-cutting skills for environment setup and computational performance.

**Skill index**: [general_data_analysis/SKILL.md](./general_data_analysis/SKILL.md)

**Skills**:
- **Environment Management**: Conda/Mamba/venv setup for reproducible environments
- **Parallel Computing**: Multi-core CPU, GPU acceleration, memory optimization

---

## Supplementary Reference: SC Best Practices

Comprehensive guidance derived from the
[Single-cell Best Practices](https://www.sc-best-practices.org) book.
Use as supplementary context when the core skills above need deeper background.

**Skill index**: [sc_best_practices/SKILL.md](./sc_best_practices/SKILL.md)

**Topics covered**:
- Preprocessing, normalization, dimensionality reduction
- Clustering, annotation, dataset integration
- Trajectory analysis, RNA velocity, lineage tracing
- Differential expression, compositional analysis, pathway analysis
- Gene regulatory networks, cell-cell communication
- Bulk deconvolution, scATAC-seq, spatial omics
- CITE-seq, immune repertoire (TCR/BCR)
- Multimodal integration, reproducibility

---

## Using Skills

1. **Before analysis**: Scan this index for relevant skills
2. **Load skill file**: Read the full skill document for detailed guidance
3. **Follow best practices**: Use the code snippets and workflows provided
4. **Adapt as needed**: Skills are templates; adjust for your specific data

讨论

基于 GitHub Discussions。登录 GitHub 即可参与讨论、点赞、订阅更新。