样本id substr和重复问题

[ShixiangWang]

@lishensuo 我发现这个数据有点问题，你有空看看能否检查和重新生成下，不是很着急。

UCSCXenaShiny/data-raw/raw-data-collection-and-tidying.R

Lines 98 to 110 in 4523aa0

	#--------purity and ploidy data-----------------------------
	# access date:2020-06-17
	# from https://gdc.cancer.gov/about-data/publications/PanCanStemness-2018
	## genome instability
	gi_data <- data.table::fread("data-raw/Purity_Ploidy_All_Samples_9_28_16.tsv", data.table = F)
	gi_data <- gi_data %>%
	dplyr::select(c(3, 5, 6, 7, 9, 10))
	gi_data <- gi_data %>%
	dplyr::select(sample, purity, ploidy, Genome_doublings = `Genome doublings`, Cancer_DNA_fraction = `Cancer DNA fraction`, Subclonal_genome_fraction = `Subclonal genome fraction`) %>%
	dplyr::mutate(sample = stringr::str_sub(sample, 1, 15))
	tcga_genome_instability <- gi_data
	attr(tcga_genome_instability, "data_source") <- "DOI:https://doi.org/10.1016/j.cell.2018.03.034"

[lishensuo]

老师，您指的问题是数据操作层面还是数据生物学含义层面？目前复现了下步骤，应该是可以得到相同的结果的

[ShixiangWang]

@lishensuo stringr::str_sub(sample, 1, 15) 这里写的有问题，并不是所有样本都是按标准 TCGA 样本 id 标注的。

[lishensuo]

收到，老师。已发现这个问题。如下是相应的解决方法。

library(dplyr)

# 部分TCGA样本ID不符合常规的格式
gi_data_raw <- data.table::fread("Purity_Ploidy_All_Samples_9_28_16.tsv", data.table = F)
gi_data <- gi_data_raw %>%
  dplyr::select(c(3, 5, 6, 7, 9, 10)) %>%
  dplyr::select(sample, purity, ploidy, Genome_doublings = `Genome doublings`, 
                Cancer_DNA_fraction = `Cancer DNA fraction`, 
                Subclonal_genome_fraction = `Subclonal genome fraction`)

head(gi_data$sample[!grepl("^TCGA", gi_data$sample)])
# [1] "GBM-TCGA-02-0001-Tumor" "GBM-TCGA-02-0006-Tumor" "GBM-TCGA-02-0007-Tumor" "GBM-TCGA-02-0009-Tumor"
# [5] "GBM-TCGA-02-0010-Tumor" "GBM-TCGA-02-0011-Tumor"

table(grepl("^TCGA", gi_data$sample))
# FALSE  TRUE 
# 795  9997 

• 9997个样本符合常规TCGA ID
• 795个样本不符合

如下想到两种解决的方式

方式1：直接丢弃795个样本

gi_data_res1 = gi_data %>% 
  dplyr::filter(grepl("^TCGA", sample)) %>%
  dplyr::mutate(sample = stringr::str_sub(sample, 1, 15))


gi_data_res1 = gi_data_res1[order(gi_data_res1$sample),]
rownames(gi_data_res1) = seq(nrow(gi_data_res1))

write.csv(gi_data_res1, file = "gi_data_res1.csv", row.names = FALSE)

方式2：将795个样本ID转为规范的ID

gi_data_out = gi_data[!grepl("^TCGA", gi_data$sample),]
table(!grepl("Tumor", gi_data_out$sample) )
# FALSE  TRUE 
# 775    20 

有两种不符合的ID类型

# (1) Tumor标记
head(gi_data_out$sample[grepl("Tumor", gi_data_out$sample)])
# [1] "GBM-TCGA-02-0001-Tumor" "GBM-TCGA-02-0006-Tumor" "GBM-TCGA-02-0007-Tumor"
# [4] "GBM-TCGA-02-0009-Tumor" "GBM-TCGA-02-0010-Tumor" "GBM-TCGA-02-0011-Tumor"

# (2) 其它类型
gi_data_out$sample[!grepl("Tumor", gi_data_out$sample)]
# [1] "WU-BRCA-TCGA-A8-A06Y-01A-21W-A019-09"  "WU-BRCA-TCGA-A8-A07R-01A-21W-A050-09" 
# [3] "WU-BRCA-TCGA-A8-A083-01A-21W-A019-09"  "WU-BRCA-TCGA-A8-A084-01A-21W-A019-09" 
# [5] "WU-BRCA-TCGA-A8-A08F-01A-11W-A019-09"  "WU-BRCA-TCGA-A8-A09E-01A-11W-A019-09" 
# [7] "WU-BRCA-TCGA-AO-A0JL-01A-11W-A071-09"  "WU-BRCA-TCGA-AO-A12B-01A-11D-A10M-09" 
# [9] "WU-BRCA-TCGA-AR-A0TS-01A-11D-A10Y-09"  "WU-BRCA-TCGA-AR-A2LL-01A-11D-A17W-09" 
# [11] "WU-BRCA-TCGA-AR-A2LR-01A-12D-A18P-09"  "WU-BRCA-TCGA-B6-A0X4-01A-11D-A10G-09" 
# [13] "WU-BRCA-TCGA-BH-A1F5-01A-12D-A13L-09"  "WU-BRCA-TCGA-D8-A146-01A-31D-A10Y-09" 
# [15] "WU-BRCA-TCGA-E9-A1RD-01A-11D-A159-09"  "BCM-KICH-TCGA-KL-8327-01A-11D-2310-10"
# [17] "BCM-KICH-TCGA-KL-8333-01A-11D-2310-10" "BCM-KICH-TCGA-KL-8342-01A-11D-2310-10"
# [19] "BCM-KICH-TCGA-KN-8419-01A-11D-2310-10" "BCM-KICH-TCGA-KN-8430-01A-11D-2310-10"

• 修改方式：均修改为TCGA的01肿瘤label

gi_data_out = gi_data_out %>% 
  dplyr::mutate(sample = stringr::str_extract(sample, 'TCGA-\\w{2}-\\w{4}')) %>% 
  dplyr::mutate(sample = paste0(sample, "-01")) %>% 
  dplyr::filter(sample != "NA-01")

gi_data_res2 = rbind(gi_data_res1, gi_data_out)
gi_data_res2 = gi_data_res2[order(gi_data_res2$sample),]
rownames(gi_data_res2) = seq(nrow(gi_data_res2))

dim(gi_data_res2)
# [1] 10791     6

write.csv(gi_data_res2, file = "gi_data_res2.csv", row.names = FALSE)

---

附件分别为两种方式的计算结果，老师觉得哪种方式更好一些呢？

gi_data_res1.csv
gi_data_res2.csv

[ShixiangWang]

采用方式1吧，目前针对其他类型的数据也没什么用。

[lishensuo]

嗯好的，已按照方式1进行了修改，并提交