以為對read.table/read.delim很熟了，誰知又掉坑里了。
我有個3萬多行的數據集，包括樣品表達量和注釋信息。大概長這樣：
本來3萬多行，可是讀進來的時候變成了1萬多行，而且read.delim和read.table減少的行數還不一樣。我用Excel打開，再另存為txt格式讀入后，數據行數變回正常的3萬多。
MP <- read.delim("combine_test.txt",sep = '\t',header = T)
MP1 <- read.table("combine_test.txt",sep = '\t',header = T)
MP2<- read.delim("new_combine_test.txt",sep = '\t',header = T)
所以我在想是不是Rstudio的問題。於是我在Linux中測試了下，發現更詭異。
MP <- read.table("combine_test2.txt",header = T,sep='\t')
dim(MP)
MP2 <- read.delim("combine_test2.txt",header = T,sep='\t')
dim(MP2)
write.table(MP,"out.txt",col.names=T,row.names=F,sep='\t',quote=F)
write.table(MP2,"out.txt",col.names=T,row.names=F,sep='\t',quote=F)
dim顯示的都是1萬多行，原樣輸出的數據卻有3萬多行！
我意識到是數據格式的問題了。用readr來試試：
MP2 <- as.data.frame(read_delim("combine_test.txt",delim = '\t'))
變回正常了。難道base R還不如tidyverse嗎？？？我在網上查了查，終於找到原因了，那就是一個quote參數的事情。
MP3 <- read.table("combine_test.txt",sep = '\t',quote = "",header = T)
MP4 <- read.delim("combine_test.txt",sep = '\t',quote = "",header = T)
關於quote參數，那個答案是這么解釋的：
Explanation: Your data has a single quote on 59th line (( pyridoxamine 5'-phosphate oxidase (predicted)). Then there is another single quote, which complements the single quote on line 59, is on line 137 (5'-hydroxyl-kinase activity...). Everything within quote will be read as a single field of data, and quotes can include the newline character also. That's why you lose the lines in between. quote = "" disables quoting altogether.
簡單理解就是我的數據里面包含了單引號''，兩個單引號之間會當成一個字段來處理，我需要提前用quote=""將字段引起來。我檢查了下，在我的KEGG的描述中確實含有引號。
如果字段字符串中本身含有雙引號""或者其他符號時，也可能出錯。為檢查這種錯誤，可以用count.fields來統計每行的字段數，如果出現NA，則說明讀入的數據有誤。
num.fields = count.fields("combine_test.txt", sep="\t")
num.fields = count.fields("combine_test.txt", sep="\t",quote = "")
貌似read.csv不會出現這種問題，因為它提前引起來了。可見read.table確實有意想不到的錯誤發生。多了解下fread和readr系列吧。