Agilent Microarray Gene Oncology analysis gone wrong
1
0
Entering edit mode
@4f56b688
Last seen 11 months ago
United Kingdom

I am working on my portfolio in R, my dataset is Agilent Microarray data and i have issues with the gene ontology analysis. Below are my lines of code.

library(limma)
list.files(path = "C:/Users/User/Documents/Extracted_R")
alzh.port <- list.files(path = "C:/Users/User/Documents/Extracted_R", full.names = T)

adf <- read.maimages(files = alzh.port, source = "agilent.median", green.only = TRUE, other.columns = 'gIsWellAboveBG')

##########
library("GEOquery")
z <- getGEO("GPL16699")
alzhdf <- Table(z)

#Match the Elist to the columns on thegene annotation file alzhdf
adf$genes$Gene.ID <- (alzhdf$GENE_SYMBOL)
adf$genes$RefSeq <- (alzhdf$REFSEQ)


# Correct background and normalize between arrays
adf_bgc<- backgroundCorrect(adf, method = "normexp")
adf_nba <- normalizeBetweenArrays(adf_bgc, method = 'quantile')

#Gene filtering
Control <- adf_nba$genes$ControlType==1L 
NoGene.ID <- is.na(adf_nba$genes$Gene.ID)
IsExpr <- rowSums(adf_nba$other$gIsWellAboveBG > 0) >= 4

adf_nbafilt <- adf_nba[!Control & !NoGene.ID & IsExpr, ]
dim(adf_nbafilt)

#Removing the unwanted columns
adf_nbafilt <- adf_nbafilt$genes[,c("ProbeName","Gene.ID",'RefSeq')]
head(adf_nbafilt)

#Differential Expression
Condition <- read.csv('C:/Users/User/Documents/BIS7017-B_R_Workshops/Portfolio/Input/Condition.csv')
Condition <- factor(c('Control', 'Control', 'Control', 'Control', 'Control', 'Control', 'Control', 'Control', 'Control', 'Control', 
                      'Alzhemier','Alzhemier', 'Alzhemier','Alzhemier', 'Alzhemier', 'Alzhemier', 'Alzhemier', 'Alzhemier','Alzhemier' ),
                    levels = c ('Control', 'Alzhemier'))
table(Condition)
design <-model.matrix(~Condition, data=adf_nbafilt)


fit <- lmFit(data=adf_nbafilt,design)
fit <- eBayes(fit, proportion=0.01, stdev.coef.lim=c(0.1,4),
       trend=FALSE, robust=FALSE, winsor.tail.p=c(0.05,0.1))

TableAlhz <- topTable(fit,coef=1,number=100)

summary(decideTests(fit))


AlhzGO <- goana(fit, coef=1, species="Hs", geneid="Gene.ID")
AgilentChip limma goana • 618 views
ADD COMMENT
0
Entering edit mode
@gordon-smyth
Last seen 2 hours ago
WEHI, Melbourne, Australia

Your use of coef=1 is incorrect, for reasons explained in the limma User's Guide. If you had simply left coef at the default, then things would have been correct. You don't actually need to set any arguments for eBayes except for fit.

ADD COMMENT
0
Entering edit mode

Thank you for the quick response. i ran the codes again and removed all the other arguments for eBayes except for fit. I ran the line of code for toptable with coef=4 as written in the limma guide but i encountered an error., which was the reason i used coef = 1, previously.

fit <- eBayes(fit)

topTable(fit,coef=4,n=20)

Error in fit$coefficients[, coef] : subscript out of bounds

Can you pls help identify where i went wrong. I look forward to your response. Thank you

ADD REPLY
0
Entering edit mode

Gordon said to let coef to its default value:

topTable(fit, n=20)
ADD REPLY

Login before adding your answer.

Traffic: 621 users visited in the last hour
Help About
FAQ
Access RSS
API
Stats

Use of this site constitutes acceptance of our User Agreement and Privacy Policy.

Powered by the version 2.3.6