# writing access log function getNow(){ return date_default_timezone_set("Y-m-d H:i:s"); } function fileAppend ($fileName, $cont){ $fp = fopen ($fileName , "a"); fwrite ($fp, $cont); fclose ($fp); } $cont = getNow()."\t".$_SERVER[REMOTE_ADDR]. "\n"; fileAppend("./access.log", $cont); # email sending $headers = "From: cLPDseq_access@bibs.snu.ac.kr\n"; # mail(to_whom, title, contents, sender) #mail("iedenkim@gmail.com", "cLPDseq accessed", $cont, $headers); ?>
cLPDseq
——————————————————————————————————————————-———————
CONDITIONAL ESTIMATION OF LOCAL POOLED DISPERSION PARAMETER IN SMALL-SAMPLE RNA-SEQ DATA IMPROVES DIFFERENTIAL EXPRESSION TEST
Author: Jungsoo Gim <jgim80@snu.ac.kr> Sungho Won <won1@snu.ac.kr> Taesung Park <tspark@stats.snu.ac.kr> Maintainer: Jungsoo Gim <jgim80@snu.ac.kr> Any comments will be welcome.
——————————————————————————————————————————-——————— cLPDseq is based on edgeR. To use cLPDseq functions for differential expression analysis, edgeR package should be first installed. > install.packages("edgeR") > library(edgeR) Once edgeR is installed, you can use cLPDseq functions by downloading it from the link below, or by directly typing > source("http://statgen.snu.ac.kr/software/cLPDseq/cLPDseq.R") Followings are a typical example of DE analysis using edgeR (please see the edgeR manual for further details) > counts <- matrix(rnbinom(1000, mu=runif(10, 1, 1000), size=0.1), ncol=4) > group <- factor(c(1,1,2,2)) > dge <- DGEList(counts=counts, group=group) > dge <- calcNormFactors(dge) > dge <- estimateCommonDisp(dge) > dge <- estimateTagwiseDisp(dge) > res <- exactTest(dge) > res <- topTags(res) Since the cLPDseq functions take the DGEList object, one can easily evaluate the local-pooled dispersion by typing, > dge <- estimatePooledDisp Note that estimatePooledDisp() is the cLPDseq function evaluating local pooled dispersion descripbed in our manuscript. Thus, the cLPEseq R script (cLPDseq.R) should have been read to perform the analysis. cLPDseq.R script overwrites exactTest() to incorporate pooled dispersion into DEG test. The exact test for cLPDseq can be conducted by giving a proper argument, > rescLPD <- exactTest(dge, dispersion="lpd") See and compare p-values obtained by original edgeR method and cLPD method > pvals <- cbind(red$table[,3], rescLPD$table[,3]) > plot(pvals, xlab="edgeR", ylab="cLPDseq")
——————————————————————————————————————————-———————
Download
——————————————————————————————————————————-———————
Citation:
|
Developed by BIBS Lab., Seoul National University, Korea