paper-figure-walkthrough

name	paper-figure-walkthrough
description	Use when the user wants to read, understand, summarize, or explain a biomedical/scientific paper in a blog-style, figure-by-figure workflow. Trigger for requests such as "按博主风格梳理", "图解论文", "逐图解读", "帮我理解这篇文章的分析逻辑", "整理供我阅读理解", especially when a PDF, paper, article, figures, single-cell/spatial transcriptomics, omics analysis, or biomedical research result is involved.

Paper Figure Walkthrough

Goal

Produce a concise Chinese-first reading guide that follows the paper's figure order. Convert the paper into a blog-style "figure caption expansion": what each figure/panel analyzes, what result it shows, and what conclusion it supports.

Prefer the user's requested style: professional, clear, easy to understand, high information density, not verbose. Keep key English technical terms and gene/cell/pathway names.

Workflow

1. Read the paper source.

   • If the input is a PDF, extract title, abstract, methods, results, discussion, figure captions, and supplementary notes if available.
   • If the input is pasted text, identify sections and figure references from the text.
   • If figures are needed and available, inspect them only when text/captions are insufficient.

2. Build the main thread before writing.

   • Identify the research background and unresolved problem.
   • Identify data sources, sample groups, and analysis purposes.
   • Identify the central cell type/gene/pathway/model.
   • Identify the paper's evidence chain: overview -> key subtype/feature -> function -> clinical relevance -> spatial validation -> mechanism -> broader validation.

3. Write in this fixed structure unless the user asks otherwise:

   • 背景解读
   • integrate 研究概述 into the second paragraph of 背景解读
   • 数据来源及获取
   • 研究结果
   • figure-by-figure sections
   • optional 核心机制总结 only when the paper's mechanism is complex or the user asks for a final synthesis

4. For each figure, use this internal pattern:

   • Figure title: 图X：...
   • Panel group heading: a content-specific heading with panel range, e.g. 研究设计与细胞图谱构建（Fig. 1A-B）
   • Short explanation:
     • what was analyzed / shown
     • key numbers, groups, genes, pathways, or cohorts
     • what the result means in the author's logic
   • Avoid a separate 小结 after every figure unless it adds real value. Prefer compact paragraph blocks under panel headings.
   • Mention relevant supplementary figures inline when they support the same claim, e.g. Supplementary Figure 6.

5. End with a compact mechanism model only when useful.

   • Use a simple arrow chain when useful:

A / disease background
        ↓
key cell/gene/pathway change
        ↓
functional phenotype
        ↓
cell communication / spatial niche / validation
        ↓
pathological meaning or therapeutic implication

Output Style

• Write primarily in Chinese.
• Keep key terms in English where they are useful: scRNA-seq, spatial transcriptomics, Fibroblast, CellChat, RCTD, FN1-ITGB1.
• When a professional English term, abbreviation, method, cell type, dataset label, score, pathway, or spatial region first appears, add a concise Chinese annotation in parentheses. Examples: snRNA-seq（single-nucleus RNA sequencing，单核 RNA 测序）, GRN（gene regulatory network，基因调控网络）, DAM（disease-associated microglia，疾病相关小胶质细胞）, LR（lesion rim，病灶边缘）. After the first annotation, reuse the short English term.
• Do not annotate standard gene symbols or protein names unless the paper itself expands them or the Chinese meaning is necessary for understanding.
• Use short paragraphs, compact tables, and dense information.
• Use "结果显示...", "提示...", "进一步说明..." to connect analysis and interpretation.
• Avoid long methodological lectures. Explain methods by their role in the analysis, not by algorithm details.
• Prefer direct, affirmative result narration over heavy critique. This style is for reading comprehension and note-taking, not peer review.
• Keep figure sections close to the original figure order and panel grouping. Do not reorganize figures into a new narrative unless necessary.
• Preserve concrete details from the author's version when available: cohort names, sample counts, group labels, pathway names, clinical endpoints, and marker genes.
• Use cautious wording for inferential analyses:
  • 提示
  • 可能
  • 支持
  • 作者推测
• Do not overemphasize critique unless the user asks for critical review. Include limitations only if they affect interpretation, usually near the end or omitted in a pure blog-style walkthrough.

Data Source Table Template

Use this compact table when source information is available. Prefer 样本类型 | 样本数量 | 具体分析用途 for blog-style output. Use dataset IDs inside sample/count cells when useful.

The 具体分析用途 column must be concrete, not generic. State what analyses were performed, on which sample/cell/tissue object, and with which method or enrichment framework when available. Include details such as key cell type identification, differential abundance, GRN/regulon analysis, subclustering, marker genes, differential expression, GSEA/KEGG/GO/GOBP enrichment, pseudotime/trajectory, cell-cell communication, spatial deconvolution, niche annotation, score calculation, clinical correlation, survival analysis, or experimental validation.

样本类型	样本数量	具体分析用途
scRNA-seq / snRNA-seq	...	构建 XX 组织/疾病与对照的单细胞/单核图谱；用 marker genes 注释主要细胞类型；比较疾病与对照的细胞比例；对关键细胞群做重聚类、差异表达、GSEA/KEGG/GO/GOBP 富集；如有则进行 GRN/regulon、拟时序轨迹、细胞通讯和 score 分组分析
空间转录组	...	划分空间区域/生态位；分析不同 niche 的细胞组成；进行空间 deconvolution 或相关性分析；定位关键细胞、基因、regulon 或 signature score；比较高低分区域的通路活性，验证单细胞发现的空间分布
bulk RNA-seq 队列	...	构建或验证关键 gene signature / cell score；比较疾病分组、分子分型或临床分层；做差异表达、GSEA/KEGG/GO 富集、免疫浸润分析、临床相关性和生存分析
实验验证队列/样本	...	用 IHC/IF/qPCR/western blot/flow/FISH 等验证关键细胞、基因、蛋白或空间共定位；连接组学推断与组织/功能证据
泛癌/外部验证队列	...	在独立疾病或多癌种队列中验证 signature、通路活性、细胞浸润、临床相关性或预后关系

Figure Section Template

**图X：简短标题**

研究设计与图谱构建（Fig. XA-B）  
研究方案：整合 XX 数据，对 XX 进行分析。图中展示 XX，并通过 marker gene/score/cluster 识别 XX。

细胞组成与组织分布（Fig. XC-E）  
展示 XX 在不同组织/样本/队列中的比例差异，结果显示 XX，提示 XX。

临床意义或机制验证（Fig. XF-H）  
在 XX 队列/实验中验证 XX，结果显示 XX，与 XX 结论一致。

Interpretation Rules

• When describing omics results, separate observation from interpretation.
  • Observation: "C2 CXCR4+ Fibroblast 具有最高 Cell Stemness Score。"
  • Interpretation: "提示该亚群可能具有更强状态可塑性。"
• When pseudotime/trajectory is used, state that it infers state continuity, not proven lineage.
• When cell communication tools are used, state that ligand-receptor relations are inferred from expression/coexpression unless experimental validation exists.
• When spatial transcriptomics is used, emphasize how it validates localization, co-occurrence, or spatial signaling niches.
• When validation experiments exist, connect them back to the omics hypothesis.
• In figure walkthrough mode, avoid repeatedly saying "这一图的意义是". Instead, let the final sentence of each panel group naturally state the implication.
• If the source paper uses grouped patient states, reproduce the group labels explicitly, e.g. low, intermediate, high, and explain them once.

Final Summary Template

**核心机制总结**

```text
研究背景/疾病状态
        ↓
关键细胞亚群/基因/通路
        ↓
功能表型
        ↓
细胞通讯或空间定位
        ↓
疾病机制或治疗启示
```

**文章最重要的结论**

1. ...
2. ...
3. ...
```