标签：
Affymetrix阵列包含一些控制探针组，用“AFFX”、“INTRON”、“EXON”注释。
一、一些芯片的控制探针组是用“AFFX”开头的，没有“INTRON”和“EXON”探针组，如“HG_U95Av2”。获取“HG_U95Av2”的“AFFX”探针组名：
library(affy)
cdfname <- "HG-U133A_2"
how = getOption("BioC")$affy$probesloc
verbose = FALSE
badOut <- list()
for (i in 1:length(how))
{
cur <- how[[i]]
envir <- switch(cur$what,
environment = cdfFromEnvironment(cdfname, cur$where, verbose),
libPath = cdfFromLibPath(cdfname, cur$where, verbose = verbose),
bioC = cdfFromBioC(cdfname, cur$where, verbose))
}
genenames <- ls(envir)
affxnames <- unique(genenames[grep("AFFX",genenames,fixed=TRUE)])
> affxnames
[1] "AFFX-BioB-3_at" "AFFX-BioB-5_at"
[3] "AFFX-BioB-M_at" "AFFX-BioC-3_at"
[5] "AFFX-BioC-5_at" "AFFX-BioDn-3_at"
[7] "AFFX-BioDn-5_at" "AFFX-CreX-3_at"
[9] "AFFX-CreX-5_at" "AFFX-DapX-3_at"
[11] "AFFX-DapX-5_at" "AFFX-DapX-M_at"
[13] "AFFX-HSAC07/X00351_3_at" "AFFX-HSAC07/X00351_5_at"
[15] "AFFX-HSAC07/X00351_M_at" "AFFX-hum_alu_at"
[17] "AFFX-HUMGAPDH/M33197_3_at" "AFFX-HUMGAPDH/M33197_5_at"
[19] "AFFX-HUMGAPDH/M33197_M_at" "AFFX-HUMISGF3A/M97935_3_at"
[21] "AFFX-HUMISGF3A/M97935_5_at" "AFFX-HUMISGF3A/M97935_MA_at"
[23] "AFFX-HUMISGF3A/M97935_MB_at" "AFFX-HUMRGE/M10098_3_at"
[25] "AFFX-HUMRGE/M10098_5_at" "AFFX-HUMRGE/M10098_M_at"
[27] "AFFX-LysX-3_at" "AFFX-LysX-5_at"
[29] "AFFX-LysX-M_at" "AFFX-M27830_3_at"
[31] "AFFX-M27830_5_at" "AFFX-M27830_M_at"
[33] "AFFX-PheX-3_at" "AFFX-PheX-5_at"
[35] "AFFX-PheX-M_at" "AFFX-r2-Bs-dap-3_at"
[37] "AFFX-r2-Bs-dap-5_at" "AFFX-r2-Bs-dap-M_at"
[39] "AFFX-r2-Bs-lys-3_at" "AFFX-r2-Bs-lys-5_at"
[41] "AFFX-r2-Bs-lys-M_at" "AFFX-r2-Bs-phe-3_at"
[43] "AFFX-r2-Bs-phe-5_at" "AFFX-r2-Bs-phe-M_at"
[45] "AFFX-r2-Bs-thr-3_s_at" "AFFX-r2-Bs-thr-5_s_at"
[47] "AFFX-r2-Bs-thr-M_s_at" "AFFX-r2-Ec-bioB-3_at"
[49] "AFFX-r2-Ec-bioB-5_at" "AFFX-r2-Ec-bioB-M_at"
[51] "AFFX-r2-Ec-bioC-3_at" "AFFX-r2-Ec-bioC-5_at"
[53] "AFFX-r2-Ec-bioD-3_at" "AFFX-r2-Ec-bioD-5_at"
[55] "AFFX-r2-P1-cre-3_at" "AFFX-r2-P1-cre-5_at"
[57] "AFFX-ThrX-3_at" "AFFX-ThrX-5_at"
[59] "AFFX-ThrX-M_at" "AFFX-TrpnX-3_at"
[61] "AFFX-TrpnX-5_at" "AFFX-TrpnX-M_at"
从affxnames的输出结果中可以看到AFFX-r2-Bs-lys-3_at、AFFX-r2-Bs-dap-3_at、AFFX-r2-Bs-phe-3_at、AFFX-r2-Bs-thr-3_s_at、AFFX-r2-Ec-bioB-3_at、AFFX-r2-Ec-bioC-3_at、AFFX-r2-Ec-bioD-3_at、AFFX-r2-P1-cre-3_at、AFFX-HSAC07/X00351_3_at、AFFX-HUMGAPDH/M33197_3_at这类探针组，它们是一些嵌入探针组或beta-actin、GAPDH基因的探针组。
二、一些芯片包含了“AFFX”、“INTRON”和“EXON”探针组，不过这些探针组也并没有用“AFFX”、“INTRON”、“EXON”显式地表示，如“HuGene-1_0-st-v1”。获取“HuGene-1_0-st-v1”的控制探针组名：
library(affy)
cdfname <- "HuGene-1_0-st-v1"
how = getOption("BioC")$affy$probesloc
verbose = FALSE
badOut <- list()
for (i in 1:length(how))
{
cur <- how[[i]]
envir <- switch(cur$what,
environment = cdfFromEnvironment(cdfname, cur$where, verbose),
libPath = cdfFromLibPath(cdfname, cur$where, verbose = verbose),
bioC = cdfFromBioC(cdfname, cur$where, verbose))
}
genenames <- ls(envir) ## 探针名
require("ArrayTools", quietly = TRUE)
annotation <- cleancdfname(cdfname, addcdf = FALSE)
dataTable <- paste(substr(annotation,1,nchar(annotation)-2),"CONTROL",sep="")
suppressWarnings(eval(parse("",-1,paste("data(",dataTable,")",sep="")))) #ArrayTools
cntrl <- NULL
try(cntrl <- get(dataTable),TRUE)
assign("affx",cntrl[grep("affx",cntrl[,2]),1])
assign("exon",cntrl[grep("exon",cntrl[,2]),1])
assign("intron",cntrl[grep("intron",cntrl[,2]),1])
controlnames <- cntrl[,1] ## 所有的控制探针组
affxnames <- get("affx")[get("affx") %in% genenames] ## “AFFX”探针组
exonnames <- get("exon")[get("exon") %in% genenames] ## “EXON”探针组
intronnames <- get("intron")[get("intron") %in% genenames] ## “INTRON”探针组
> affxnames
[1] 7892796 7892925 7893130 7893306 7893613 7893939 7894584 7894611 7894970
[10] 7895139 7895220 7896160 7896366
标签：
原文地址：http://www.cnblogs.com/xianwen/p/4522370.html
