graph TB;
subgraph Phase I: Data Collection & Pre-processing
direction TB;
Data_SC["<b>scRNA-seq Datasets</b><br/>(GSE datasets)<br/><i>Identification of cell types</i>"];
Data_Bulk["<b>Bulk Transcriptome</b><br/>Training: TCGA-gastric cancer<br/>Validation: GEO Cohorts"];
Data_Multi["<b>Multi-omics Data</b><br/>Mutation, CNV, Methylation<br/>HPA (IHC), Drug Databases"];
end
subgraph Phase II: Identification of FRGs
direction TB;
Step_Clustering["Seurat Clustering & SingleR Annotation<br/>(Immune, Stromal, Malignant cells)"];
Step_Screening["Screening Fibroblast-Related Genes (FRGs)<br/>High expression in fibroblast clusters"];
Step_Validation["Cross-Validation<br/>(TISCH DB & Correlation with<br/>xCell/MCP-counter)"];
Node_54FRGs{"<b>Candidate FRGs</b><br/>Identified"};
end
subgraph Phase III: FRGI Model Construction
direction TB;
Step_Omics["Multi-omics Characterization<br/>(Diff. Exp, Mutation, CNV, Methylation)"];
Step_Drug["Clinical Value Assessment<br/>(PubTator, TDL, Gene Dependency Score)"];
Step_ML["<b>Machine Learning Selection</b><br/>10 Algorithms Integration<br/>(Lasso, Ridge, RF, GBM, Boruta, etc.)"];
Node_3Genes{"<b>Core Signature (FRGI)</b>"};
end
subgraph Phase IV: CD8+ T-cell - Fibroblast Subtyping
direction TB;
Step_Class1["<b>FRG Subtyping</b><br/>Fibroblast Hot vs. Cold"];
Step_Mechanism["Mechanism Analysis<br/>GSEA, GSVA, TIDE (Immunotherapy)"];
Step_Class2["<b>Dual-Factor Subtyping<br/>Construction</b><br/>CD8+ T-cell Activity <b>x</b> FRG Status"];
Outcome_4Types["<b>Four Molecular Subtypes</b><br/>CD8+FH, CD8+FC, CD8-FH, CD8-FC"];
end
Step_Nomogram["Nomogram Construction<br/>FRGI + Clinical Variables"];
Step_SurvVal["Survival Validation<br/>(TCGA, GEO, Pan-cancer)"];
Outcome_Therapy["<b>Therapeutic Strategy Map</b><br/>Precision Medicine Guidance<br/>(Immuno/Chemo/Targeted Therapy)"];
Data_SC --> Step_Clustering;
Step_Clustering --> Step_Screening;
Step_Screening --> Step_Validation;
Step_Validation --> Node_54FRGs;
Node_54FRGs --> Step_Omics;
Node_54FRGs --> Step_Drug;
Data_Multi --> Step_Omics;
Step_Omics --> Step_ML;
Step_Drug --> Step_ML;
Data_Bulk -- "Expression Data" --> Step_ML;
Step_ML --> Node_3Genes;
Node_3Genes --> Step_Nomogram;
Node_3Genes --> Step_SurvVal;
Node_3Genes --> Step_Class1;
Step_Class1 --> Step_Mechanism;
Step_Mechanism -- "Integrate with<br/>CD8+ T-cell Signature" --> Step_Class2;
Step_Class2 --> Outcome_4Types;
Outcome_4Types --> Outcome_Therapy;
Step_Nomogram -. "Clinical Prediction" .-> Outcome_Therapy;
Step_SurvVal --> Outcome_Therapy;
%% Styling
style Data_SC fill:#E3F2FD,stroke:#1565C0,stroke-width:1.5px;
style Data_Bulk fill:#E3F2FD,stroke:#1565C0,stroke-width:1.5px;
style Data_Multi fill:#E3F2FD,stroke:#1565C0,stroke-width:1.5px;
style Step_Clustering fill:#E0F2F1,stroke:#00695C,stroke-width:1.5px;
style Step_Screening fill:#E0F2F1,stroke:#00695C,stroke-width:1.5px;
style Step_Validation fill:#E0F2F1,stroke:#00695C,stroke-width:1.5px;
style Node_54FRGs fill:#B2DFDB,stroke:#00695C,stroke-width:1.5px;
style Step_Omics fill:#FFF3E0,stroke:#EF6C00,stroke-width:1.5px;
style Step_Drug fill:#FFF3E0,stroke:#EF6C00,stroke-width:1.5px;
style Step_ML fill:#FFF3E0,stroke:#EF6C00,stroke-width:1.5px;
style Node_3Genes fill:#FFE0B2,stroke:#EF6C00,stroke-width:1.5px;
style Step_Nomogram fill:#F3E5F5,stroke:#6A1B9A,stroke-width:1.5px;
style Step_SurvVal fill:#F3E5F5,stroke:#6A1B9A,stroke-width:1.5px;
style Step_Class1 fill:#FFEBEE,stroke:#C62828,stroke-width:1.5px;
style Step_Mechanism fill:#FFEBEE,stroke:#C62828,stroke-width:1.5px;
style Step_Class2 fill:#FFEBEE,stroke:#C62828,stroke-width:2.5px;
style Outcome_4Types fill:#FAFAFA,stroke:#333333,stroke-width:1.5px;
style Outcome_Therapy fill:#FAFAFA,stroke:#333333,stroke-width:1.5px;
easybio一个简化、可交互的单细胞数据注释R包
手动注释: 耗时、主观、可复现性差。
自动化工具: 效率高,但通常是“黑盒”,结果难以验证,缺乏可信度。
如何平衡效率与可靠性,是单细胞注释的核心挑战。
easybio 解决方案:透明化三步工作流自动化匹配 (matchCellMarker2)
交互式验证 (check_marker & plotSeuratDot)
手动确认 (finsert)
整个流程可通过
|>管道符一气呵成,代码直观、逻辑清晰。
透明可信
灵活易用
现代规范
litedown 构建项目Vignette,提供轻量、现代化的使用文档。