语雀：左手柳叶刀右手炭火烧
微信公众号：研平方 | 简书：研平方
关注可了解更多的科研教程及技巧。如有问题或建议，请留言。
欢迎关注我：一起学习，一起进步！

最近，小编“扫荡”文献时，发现一个令我十分感兴趣的应用，利用文本文本挖掘技术可以评估选定基因与癌症之间的关联。提到文本挖掘这类技术，小编当然要一探究竟了。

1.原文如下

Literature evidence for the identified target genes in cancer

We used OncoScore, a text mining tool to assess the associations between each gene and specific cancers based on the literature. A cutoff value of 21.09 was suggested to determine true positives and the true negatives in cancer gene identification.

2.查找资料

习惯性的打开浏览器，准备打破砂锅问到底，惊喜的发现，OncoScore竟然是一个写好的R包，而且放在了Bioconductor网页，可直接进行安装、使用。虽然文章发在了Sci Rep杂志上，但是小编认为还是值得一试。

【生信分析】-文本挖掘目标基因 评估致癌能力，你造嘛？ - 图1

【生信分析】-文本挖掘目标基因 评估致癌能力，你造嘛？ - 图2

3.它能干什么

The OncoScore analysis consists of two parts. One can estimate a score to asses the
oncogenic potential of a set of genes, given the lecterature knowledge, at the time of the
analysis, or one can study the trend of such score over time.

可见，OncoScore不仅可以依据文献中的知识，对一组设定目标基因列表的致癌能力进行评分，还可以研究这个分数随时间的趋势。

4.开始表演，拿好小板凳看戏

4.1 准备工作

  1. if (!requireNamespace("BiocManager", quietly = TRUE))
  2.     install.packages("BiocManager")
  4. BiocManager::install("OncoScore")
  6. # load the library
  7. library(OncoScore)
  8. # Define a query
  9. query = perform.query(c("ASXL1","IDH1","IDH2","SETBP1","TET2"))
  11. ### Starting the queries for the selected genes.
  13. ### Performing queries for cancer literature 
  14.     Number of papers found in PubMed for ASXL1 was: 923 
  15.     Number of papers found in PubMed for IDH1 was: 3691 
  16.     Number of papers found in PubMed for IDH2 was: 1318 
  17.     Number of papers found in PubMed for SETBP1 was: 177 
  18.     Number of papers found in PubMed for TET2 was: 1609 
  20. ### Performing queries for all the literature 
  21.     Number of papers found in PubMed for ASXL1 was: 1018 
  22.     Number of papers found in PubMed for IDH1 was: 3902 
  23.     Number of papers found in PubMed for IDH2 was: 1499 
  24.     Number of papers found in PubMed for SETBP1 was: 229 
  25.     Number of papers found in PubMed for TET2 was: 2117

以上我们可以发现，通过检索，得到了癌症相关研究的文献数量，以及所有与检索基因相关文献数量。

OncoScore provides a function to merge gene names if requested by the user. This function is useful when there are aliases in the gene list.

  1. combine.query.results(query, c('IDH1', 'IDH2'), 'new_gene')
  2.          CitationsGene CitationsGeneInCancer
  3. ASXL1             1018                   923
  4. SETBP1             229                   177
  5. TET2              2117                  1609
  6. new_gene          5401                  5009

当然，OncoScore还可以依据染色体信息检索基因。这里不再演示。

4.2 重点来啦

4.2.1 开始计算基因的致癌评分
  1. result = compute.oncoscore(query)
  3. ### Processing data
  4. ### Computing frequencies scores 
  5. ### Estimating oncogenes
  6. ### Results:
  7.      ASXL1 -> 81.59349 
  8.      IDH1 -> 86.66355 
  9.      IDH2 -> 79.59096 
  10.      SETBP1 -> 67.43283 
  11.      TET2 -> 69.12424

4.2.2 时间趋势分析（OncoScore timeline analysis）
  1. query.timepoints = perform.query.timeseries(c("ASXL1","IDH1","IDH2","SETBP1","TET2"),
  2.                                             c("2012/03/01", "2013/03/01", "2014/03/01", "2015/03/01", "2016/03/01"))
  4. ### Starting the queries for the selected genes.
  5. ### Quering PubMed for timepoint 2012/03/01 
  6.     ### Performing queries for cancer literature 
  7.     Number of papers found in PubMed for ASXL1 was: 86 
  8.     Number of papers found in PubMed for IDH1 was: 409 
  9.     Number of papers found in PubMed for IDH2 was: 173 
  10.     Number of papers found in PubMed for SETBP1 was: 5 
  11.     Number of papers found in PubMed for TET2 was: 173 
  12.     ### Performing queries for all the literature 
  13.     Number of papers found in PubMed for ASXL1 was: 92 
  14.     Number of papers found in PubMed for IDH1 was: 489 
  15.     Number of papers found in PubMed for IDH2 was: 235 
  16.     Number of papers found in PubMed for SETBP1 was: 10 
  17.     Number of papers found in PubMed for TET2 was: 197 
  18. ### Quering PubMed for timepoint 2013/03/01 
  19.     ### Performing queries for cancer literature 
  20.     Number of papers found in PubMed for ASXL1 was: 135 
  21.     Number of papers found in PubMed for IDH1 was: 662 
  22.     Number of papers found in PubMed for IDH2 was: 267 
  23.     Number of papers found in PubMed for SETBP1 was: 11 
  24.     Number of papers found in PubMed for TET2 was: 258 
  25.     ### Performing queries for all the literature 
  26.     Number of papers found in PubMed for ASXL1 was: 150 
  27.     Number of papers found in PubMed for IDH1 was: 753 
  28.     Number of papers found in PubMed for IDH2 was: 336 
  29.     Number of papers found in PubMed for SETBP1 was: 18 
  30.     Number of papers found in PubMed for TET2 was: 303 
  31. ### Quering PubMed for timepoint 2014/03/01 
  32.     ### Performing queries for cancer literature 
  33.     Number of papers found in PubMed for ASXL1 was: 188 
  34.     Number of papers found in PubMed for IDH1 was: 904 
  35.     Number of papers found in PubMed for IDH2 was: 365 
  36.     Number of papers found in PubMed for SETBP1 was: 29 
  37.     Number of papers found in PubMed for TET2 was: 347
  38.     ### Performing queries for all the literature 
  39.     Number of papers found in PubMed for ASXL1 was: 209 
  40.     Number of papers found in PubMed for IDH1 was: 1003 
  41.     Number of papers found in PubMed for IDH2 was: 440 
  42.     Number of papers found in PubMed for SETBP1 was: 36 
  43.     Number of papers found in PubMed for TET2 was: 431 
  44. ### Quering PubMed for timepoint 2015/03/01 
  45.     ### Performing queries for cancer literature 
  46.     Number of papers found in PubMed for ASXL1 was: 257 
  47.     Number of papers found in PubMed for IDH1 was: 1198 
  48.     Number of papers found in PubMed for IDH2 was: 468 
  49.     Number of papers found in PubMed for SETBP1 was: 51 
  50.     Number of papers found in PubMed for TET2 was: 461 
  51.     ### Performing queries for all the literature 
  52.     Number of papers found in PubMed for ASXL1 was: 286 
  53.     Number of papers found in PubMed for IDH1 was: 1304 
  54.     Number of papers found in PubMed for IDH2 was: 551 
  55.     Number of papers found in PubMed for SETBP1 was: 66 
  56.     Number of papers found in PubMed for TET2 was: 583 
  57. ### Quering PubMed for timepoint 2016/03/01 
  58.     ### Performing queries for cancer literature 
  59.     Number of papers found in PubMed for ASXL1 was: 323 
  60.     Number of papers found in PubMed for IDH1 was: 1506 
  61.     Number of papers found in PubMed for IDH2 was: 569 
  62.     Number of papers found in PubMed for SETBP1 was: 68 
  63.     Number of papers found in PubMed for TET2 was: 587
  64.     ### Performing queries for all the literature 
  65.     Number of papers found in PubMed for ASXL1 was: 359 
  66.     Number of papers found in PubMed for IDH1 was: 1625 
  67.     Number of papers found in PubMed for IDH2 was: 661 
  68.     Number of papers found in PubMed for SETBP1 was: 89 
  69.     Number of papers found in PubMed for TET2 was: 745

perform.query.timeseries ()函数检索了几个设定时间的文献数据信息。

  1. result.timeseries = compute.oncoscore.timeseries(query.timepoints)
  3. ### Computing oncoscore for timepoint 2012/03/01 
  4. ### Processing data
  5. ### Computing frequencies scores 
  6. ### Estimating oncogenes
  7. ### Results:
  8.      ASXL1 -> 79.14893 
  9.      IDH1 -> 74.27776 
  10.      IDH2 -> 64.27063 
  11.      SETBP1 -> 34.9485 
  12.      TET2 -> 76.29579 
  13. ### Computing oncoscore for timepoint 2013/03/01 
  14. ### Processing data
  15. ### Computing frequencies scores 
  16. ### Estimating oncogenes
  17. ### Results:
  18.      ASXL1 -> 77.54983 
  19.      IDH1 -> 78.71551 
  20.      IDH2 -> 69.99559 
  21.      SETBP1 -> 46.4559 
  22.      TET2 -> 74.81894 
  23. ### Computing oncoscore for timepoint 2014/03/01 
  24. ### Processing data
  25. ### Computing frequencies scores 
  26. ### Estimating oncogenes
  27. ### Results:
  28.      ASXL1 -> 78.28121 
  29.      IDH1 -> 81.08963 
  30.      IDH2 -> 73.50788 
  31.      SETBP1 -> 64.97398 
  32.      TET2 -> 71.31087 
  33.      ### Computing oncoscore for timepoint 2015/03/01 
  34. ### Processing data
  35. ### Computing frequencies scores 
  36. ### Estimating oncogenes
  37. ### Results:
  38.      ASXL1 -> 78.84769 
  39.      IDH1 -> 82.99363 
  40.      IDH2 -> 75.60886 
  41.      SETBP1 -> 64.48853 
  42.      TET2 -> 70.46695 
  43. ### Computing oncoscore for timepoint 2016/03/01 
  44. ### Processing data
  45. ### Computing frequencies scores 
  46. ### Estimating oncogenes
  47. ### Results:
  48.      ASXL1 -> 79.37202 
  49.      IDH1 -> 83.9881 
  50.      IDH2 -> 76.89328 
  51.      SETBP1 -> 64.60591 
  52.      TET2 -> 70.53378

4.2.3 可视化
  1. ## Oncogenetic potential of the considered genes
  2. plot.oncoscore(result, col = 'darkblue')
  4. ## Absolute values of the oncogenetic potential of the considered genes over times
  5. plot.oncoscore.timeseries(result.timeseries)
  7. ## Variations of the oncogenetic potential of the considered genes over times
  8. plot.oncoscore.timeseries(result.timeseries,
  9.                           incremental = TRUE,
  10.                           ylab='absolute variation')
  12. ## Variations as relative values of the oncogenetic potential of the considered genes over times
  13. plot.oncoscore.timeseries(result.timeseries,
  14.                           incremental = TRUE,
  15.                           relative = TRUE,
  16.                           ylab='relative variation')

【生信分析】-文本挖掘目标基因 评估致癌能力，你造嘛？ - 图3

【生信分析】-文本挖掘目标基因 评估致癌能力，你造嘛？ - 图4

【生信分析】-文本挖掘目标基因 评估致癌能力，你造嘛？ - 图5

【生信分析】-文本挖掘目标基因 评估致癌能力，你造嘛？ - 图6
Using the ONCOSCORE package.pdf