2. # luad
    3. # 导入数据，并整理
    4. rm(list = ls())
    5. load( file = "./Rawdata/TCGA_LUNG_id_exp_sur.RData")
    7. # 提取突变数据
    8. d1 = as.data.frame(luad_mc3@data)
    9. d2 = d1[,c("Hugo_Symbol","Tumor_Sample_Barcode")]
    10. class(d2)
    12. a = as.data.frame(table(d2$Tumor_Sample_Barcode))
    13. a = as.character(a[,1])
    15. # 提取突变基因
    16. k = c("ALK")
    17. d3 = d2[d2$Hugo_Symbol %in% k, ]
    18. muid = d3$Tumor_Sample_Barcode
    19. muid = muid[!duplicated(muid)]
    20. muid = as.character(muid)
    22. # 整理突变病人ID
    23. library(stringr)
    24. muid = str_split(muid, "-", simplify = T)[,c(1, 2, 3)]
    26. muid2 = rnorm(30)
    27. for(i in 1:nrow(muid)) {
    28.   muid2[i] = paste(c(muid[i,], "01"), sep = "-", collapse = "-")
    29. }
    30. muid2[14] = "TCGA-50-5946-02"
    31. muid2 = muid2[muid2 %in% LUNG_survival_data$sample]
    33. luad.alk.mu.id = muid2
    36. ## lusc
    37. # 导入数据，并整理
    38. load( file = "./Rawdata/TCGA_LUNG_id_exp_sur.RData")
    40. # 提取突变数据
    41. d1 = as.data.frame(lusc_mc3@data)
    42. d2 = d1[,c("Hugo_Symbol","Tumor_Sample_Barcode")]
    43. class(d2)
    45. a = as.data.frame(table(d2$Tumor_Sample_Barcode))
    46. a = as.character(a[,1])
    48. # 提取突变基因
    49. k = c("ALK")
    50. d3 = d2[d2$Hugo_Symbol %in% k, ]
    51. muid = d3$Tumor_Sample_Barcode
    52. muid = muid[!duplicated(muid)]
    53. muid = as.character(muid)
    55. # 整理突变病人ID
    56. library(stringr)
    57. muid = str_split(muid, "-", simplify = T)[,c(1, 2, 3)]
    59. muid2 = rnorm(30)
    60. for(i in 1:nrow(muid)) {
    61.   muid2[i] = paste(c(muid[i,], "01"), sep = "-", collapse = "-")
    62. }
    64. muid2 = muid2[muid2 %in% LUNG_survival_data$sample]
    66. lusc.alk.mu.id = muid2
    68. save(luad.alk.mu.id, lusc.alk.mu.id, file = "./Rdata/luad.lusc.alk.muid.Rdata")