Scholargraph Fusion Genes

# 1. 进入 ScholarGraph 目录
cd /root/.openclaw/workspace/skills/ScholarGraph

# 2. 配置 MiniMax API Key (或使用其他 provider)
export AI_PROVIDER=minimax
export MINIMAX_API_KEY="your-api-key"

# 3. 验证配置
~/.bun/bin/bun run cli.ts --help

# 创建工作目录
mkdir -p /workspace/research/fusion/review
mkdir -p /workspace/research/fusion/reports

# 第一轮: 搜索各癌种融合基因 review
~/.bun/bin/bun run cli.ts search "lung cancer fusion gene review" --limit 10
~/.bun/bin/bun run cli.ts search "breast cancer fusion gene review" --limit 10
~/.bun/bin/bun run cli.ts search "colorectal cancer fusion gene review" --limit 10
~/.bun/bin/bun run cli.ts search "solid tumor fusion gene review" --limit 10

# 搜索特定基因的融合信息 (并行执行)
~/.bun/bin/bun run cli.ts search "EML4-ALK fusion variant V1 V2 V3 breakpoint" --limit 10
~/.bun/bin/bun run cli.ts search "ROS1 fusion partner CD74 SDC4 TPM3" --limit 10
~/.bun/bin/bun run cli.ts search "RET fusion KIF5B CCDC6 NCOA4 breakpoint" --limit 10
~/.bun/bin/bun run cli.ts search "NTRK1 NTRK2 NTRK3 fusion partner ETV6" --limit 10
~/.bun/bin/bun run cli.ts search "FGFR fusion partner TACC3 BICC1" --limit 10
~/.bun/bin/bun run cli.ts search "NRG1 fusion CD74 SDC4 cancer" --limit 10

# 使用 Europe PMC 下载免费文献 (推荐)
# 搜索 PMID 后访问: https://europepmc.org/articles/PMC[ID]?pdf=render
curl -L -o "/workspace/research/fusion/review/filename.pdf" "https://europepmc.org/articles/PMC[ID]?pdf=render"

# 或使用 ScholarGraph download (需配置 SERPER_API_KEY)
~/.bun/bin/bun run cli.ts download "RET fusion cancer review" --limit 3 --output /workspace/research/fusion/review

# 补充搜索可能遗漏的融合基因
~/.bun/bin/bun run cli.ts search "EGFRvIII glioma fusion deletion" --limit 5
~/.bun/bin/bun run cli.ts search "ESR1 fusion breast cancer endocrine resistance" --limit 5
~/.bun/bin/bun run cli.ts search "BRAF fusion KIAA1549 melanoma" --limit 5
~/.bun/bin/bun run cli.ts search "MET fusion CAPZA2 LAYN lung cancer" --limit 5
~/.bun/bin/bun run cli.ts search "ALK fusion HIP1 KIF5B non-EML4" --limit 5

// merge_results.js
const fs = require('fs');

// 假设有两轮搜索结果
const round1 = require('./round1.json');
const round2 = require('./round2.json');

// 合并去重函数
function mergeAndDeduplicate(results1, results2) {
  const seen = new Map();
  const merged = [];
  
  for (const item of [...results1, ...results2]) {
    // 根据基因+融合伙伴作为唯一标识
    const key = `${item.gene}-${item.fusion}`;
    if (!seen.has(key)) {
      seen.set(key, true);
      merged.push(item);
    }
  }
  return merged;
}

const merged = mergeAndDeduplicate(round1, round2);
fs.writeFileSync('./merged_results.json', JSON.stringify(merged, null, 2));
console.log(`Merged: ${merged.length} unique records`);

# 安装 xlsx 库
npm install xlsx

const XLSX = require('xlsx');

// ========== 辅助函数 ==========

// 频率级别标签逻辑
function getFreqLabel(freq) {
  // 特殊情况: EGFR扩增等
  if (freq.includes('EGFR扩增')) return '-';
  
  // 提取所有数字
  const nums = freq.match(/\d+/g)?.map(Number) || [];
  if (nums.length === 0) return '-';
  
  // 取最大值进行判断
  const maxVal = Math.max(...nums);
  
  // 频率级别判断
  if (maxVal >= 25) return '高';    // ≥25%: 高频率
  if (maxVal >= 10) return '中';    // 10-24%: 中频率
  if (maxVal >= 3) return '低';     // 3-9%: 低频率
  return '极低';                     // <3%: 极低频率
}

// 参考文献数量统计
function countSources(ref) {
  if (!ref) return 0;
  const sources = ref.split(/;\s*/).filter(s => s.trim().length > 0);
  return sources.length;
}

// 标准转录本映射
const transcriptMap = {
  'ALK': 'NM_004935',
  'ROS1': 'NM_002944',
  'RET': 'NM_020975',
  'NTRK1': 'NM_002529',
  'NTRK2': 'NM_006180',
  'NTRK3': 'NM_002449',
  'MET': 'NM_000245',
  'FGFR1': 'NM_023110',
  'FGFR2': 'NM_000141',
  'FGFR3': 'NM_001142',
  'NRG1': 'NM_013962',
  'BRAF': 'NM_004333',
  'EGFR': 'NM_005228',
  'ESR1': 'NM_000125',
};

// ============================

// 完整数据格式 - 每行独立完整
// 字段: 基因, 融合类型, 变体, 融合转录本, 参考基因组, 5'基因, 5'基因断裂位置, 3'基因, 3'基因断裂位置, 频率, 癌种, 临床意义, 靶向药物, 参考文献
const fusionData = [
  // 示例:
  { 
    "基因": "ALK", 
    "融合类型": "EML4-ALK", 
    "变体": "V1",
    "融合转录本": "NM_004935",
    "参考基因组": "GRCh37",
    "5'基因": "EML4", 
    "5'基因断裂位置": "exon 13", 
    "3'基因": "ALK", 
    "3'基因断裂位置": "exon 20",
    "频率": "~33%", 
    "癌种": "肺癌", 
    "临床意义": "标准型", 
    "靶向药物": "Crizotinib; Alectinib", 
    "参考文献": "Zhang SS, et al. Going beneath the tip of the iceberg... Lung Cancer. 2021; PMID: 34175504"
  },
  // ... 更多记录
];

// 添加频率级别和参考文献数
const enrichedData = fusionData.map(item => ({
  ...item,
  "频率级别": getFreqLabel(item["频率"]),
  "参考文献数": countSources(item["参考文献"])
}));

const wb = XLSX.utils.book_new();

// Sheet 1: 融合基因汇总
const mainSheet = XLSX.utils.json_to_sheet(enrichedData);
XLSX.utils.book_append_sheet(wb, mainSheet, "融合基因汇总");

// Sheet 2: 探针设计推荐
const probeData = [
  { "基因": "ALK", "DNA探针区域": "intron 19", "RNA探针区域": "exon 18-21", "探针长度": "300-500bp", "备注": "覆盖所有变体" },
  { "基因": "ROS1", "DNA探针区域": "intron 31-35", "RNA探针区域": "exon 32-36", "探针长度": "400-600bp", "备注": "覆盖常见融合伙伴" },
  // ...
];
const probeSheet = XLSX.utils.json_to_sheet(probeData);
XLSX.utils.book_append_sheet(wb, probeSheet, "探针设计推荐");

// Sheet 3: 分癌种汇总
const cancerSummary = [
  { "癌种": "肺癌", "融合数": 15, "主要基因": "ALK, ROS1, RET, MET" },
  { "癌种": "乳腺癌", "融合数": 3, "主要基因": "ESR1, NTRK" },
  // ...
];
const cancerSheet = XLSX.utils.json_to_sheet(cancerSummary);
XLSX.utils.book_append_sheet(wb, cancerSheet, "分癌种汇总");

// 保存文件
XLSX.writeFile(wb, '/workspace/research/fusion/reports/fusion_genes_report.xlsx');

node generate_excel.js

// 频率级别判断函数
function getFreqLabel(freq) {
  // 特殊情况: EGFR扩增等
  if (freq.includes('EGFR扩增')) return '-';
  
  // 提取所有数字
  const nums = freq.match(/\d+/g)?.map(Number) || [];
  if (nums.length === 0) return '-';
  
  // 取最大值进行判断
  const maxVal = Math.max(...nums);
  
  // 频率级别判断
  if (maxVal >= 25) return '高';    // ≥25%: 高频率
  if (maxVal >= 10) return '中';    // 10-24%: 中频率
  if (maxVal >= 3) return '低';     // 3-9%: 低频率
  return '极低';                     // <3%: 极低频率
}

参考文献: Zhang SS, et al. Going beneath the tip of the iceberg: Identifying and understanding EML4-ALK variants and TP53 mutations. Lung Cancer. 2021; PMID: 34175504; Christopoulos P, et al. EML4-ALK fusion variant V3 is a high-risk feature. Clin Cancer Res. 2018

参考文献: 同上
参考文献: 同V3a

AI_PROVIDER=minimax
MINIMAX_API_KEY=sk-cp-xxx
MINIMAX_BASE_URL=https://api.minimaxi.com/v1
MINIMAX_MODEL=MiniMax-M2.5