as.vector(x)中的错误:没有将此S4类胁迫到向量的方法

Question

我试图在bash命令行上运行一个R脚本(我使用的是CentOS 8)，命令是：cat 1_myScript.R | R --slave --args $SAMPLE"_x" $SAMPLE"_y"

其中$SAMPLE是我在R脚本中指定的参数，如下所示

> args<-commandArgs()

> aaa<-args[4]

这个语法总是适用于我的所有脚本，但是现在它给了我以下错误：

as.vector(x)中的错误:没有强制此S4类执行向量调用的方法: setdiff -> setdiff.default -> -> as.vector执行暂停

奇怪的是，如果我试图在R控制台中运行这个脚本

>source("1_myScript.R")

它没有任何错误。我已经查过了，它似乎是链接到我在脚本中使用的库"GenomicRanges“的一个特性。下面是我的脚本的正文(请注意，我不知道它失败的确切行)：

#!/usr/bin/env Rscript

library(rtracklayer)
library(data.table)
library(tidyverse)
#args <- commandArgs()
#aaa<-args[4]
aaa<-gsub("_T.finalSorted.bam_CNVs","",args[1])
targDir<-"/srv/ngsdata/dalteriog/SV_analysis/example/WGS_NB_novogene/CNVDir/"
dataTable <-fread((paste0(targDir,args[2]), header=TRUE)
ratio<-data.frame(dataTable)

dataTable <-fread(paste0(targDir,args[1]), header=FALSE)
cnvs<- data.frame(dataTable)

ratio$Ratio[which(ratio$Ratio==-1)]=NA

cnvs.bed=GRanges(cnvs[,1],IRanges(cnvs[,2],cnvs[,3])) # primo ogetto del GRange obj --> chr; secondo --> range
ratio.bed=GRanges(ratio$Chromosome,IRanges(ratio$Start,ratio$Start),score=ratio$Ratio) # score è un metadata

overlaps <- subsetByOverlaps(ratio.bed,cnvs.bed) # regioni overlappanti i due df
normals <- setdiff(ratio.bed,cnvs.bed) # regioni diverse
normals <- subsetByOverlaps(ratio.bed,normals) # la stessa cosa, ma con lo score associato

#mu <- mean(score(normals),na.rm=TRUE)
#sigma<- sd(score(normals),na.rm=TRUE)

#hist(score(normals),n=500,xlim=c(0,2))
#hist(log(score(normals)),n=500,xlim=c(-1,1))

#shapiro.test(score(normals)[which(!is.na(score(normals)))][5001:10000])
#qqnorm (score(normals)[which(!is.na(score(normals)))],ylim=(c(0,10)))
#qqline(score(normals)[which(!is.na(score(normals)))], col = 2)

#shapiro.test(log(score(normals))[which(!is.na(score(normals)))][5001:10000])
#qqnorm (log(score(normals))[which(!is.na(score(normals)))],ylim=(c(-6,10)))
#qqline(log(score(normals))[which(!is.na(score(normals)))], col = 2)

numberOfCol=length(cnvs)

for (i in c(1:length(cnvs[,1]))) {
  values <- score(subsetByOverlaps(ratio.bed,cnvs.bed[i])) #score bayesiano della iesima CNV che overlappa con il file ratio 
  #wilcox.test(values,mu=mu)
  W <- function(values,normals){resultw <- try(wilcox.test(values,score(normals)), silent = TRUE)
    if(class(resultw)=="try-error") return(list("statistic"=NA,"parameter"=NA,"p.value"=NA,"null.value"=NA,"alternative"=NA,"method"=NA,"data.name"=NA)) else resultw}
  KS <- function(values,normals){resultks <- try(ks.test(values,score(normals)), silent = TRUE)
    if(class(resultks)=="try-error") return(list("statistic"=NA,"p.value"=NA,"alternative"=NA,"method"=NA,"data.name"=NA)) else resultks}
  #resultks <- try(KS <- ks.test(values,score(normals)), silent = TRUE)
  # if(class(resultks)=="try-error") NA) else resultks
  cnvs[i,numberOfCol+1]=W(values,normals)$p.value
  cnvs[i,numberOfCol+2]=KS(values,normals)$p.value
  }

if (numberOfCol==5) {
  names(cnvs)=c("chr","start","end","copy number","status","WilcoxonRankSumTestPvalue","KolmogorovSmirnovPvalue")  
}
if (numberOfCol==7) {
  names(cnvs)=c("chr","start","end","copy number","status","genotype","uncertainty","WilcoxonRankSumTestPvalue","KolmogorovSmirnovPvalue")  
}
if (numberOfCol==9) {
  names(cnvs)=c("chr","start","end","copy number","status","genotype","uncertainty","somatic/germline","precentageOfGermline","WilcoxonRankSumTestPvalue","KolmogorovSmirnovPvalue")  
}

cnvs$Wfdr <- p.adjust(cnvs$WilcoxonRankSumTestPvalue, method="BH",n=nrow(cnvs))
cnvs$KSfdr <- p.adjust(cnvs$KolmogorovSmirnovPvalue, method="BH",n=nrow(cnvs))
cnvs<-subset(cnvs, Wfdr <= 0.05 & KSfdr <= 0.05)



samp<-gsub("_T.finalSorted.bam_CNVs","",aaa)
cnvs<-add_column(cnvs, Sample=samp, .before=1)
write.table(cnvs, file=paste(targDir,aaa,"CNV.p.filtered.txt",sep=""),sep="\t",quote=F,row.names=F)

Answer 1

看来问题就在第22行：

normals <- setdiff(ratio.bed,cnvs.bed)

在本例中，我们希望找到两个granges对象之间的差异，但是执行此操作的函数是setdiff of GenomicRanges，它被包基所掩盖。答案的解决方案是这样编辑这样的行：

normals <- GenomicRanges::setdiff(ratio.bed,cnvs.bed)

再次感谢您的努力！