Get confused about the mean log-intensity (Amean)
1
0
Entering edit mode
OE • 0
@1b6d9c0f
Last seen 21 months ago
Morocco

Dear community,

I have been reading the usersguide of limma package in order to get some useful indications about how can I define the Amean's cutoff but after following the indications provided there I got a little bit confused.

In fact, in the page 34 of the usersguide it is written: " A histogram of the Amean values and a sigma vs Amean plot may help identify a cutoff below which Amean values can be filtered"

But my problem lies in the plots, I can't figure out actually how I can define an appropriate cutoff for the Amean because they get me confused.

So far the only way I use to define my Amean cutoff is through: mean(fit$Amean).

Is that true ?

Thanks in advance !

limma • 563 views
ADD COMMENT
0
Entering edit mode
@gordon-smyth
Last seen 7 hours ago
WEHI, Melbourne, Australia

the only way I use to define my Amean cutoff is through: mean(fit$Amean)

That is a potentially poor cutoff and it conflicts with the advice given to you by me (How many features an ExpressionSet can contain ?) and by the limma User's Guide.

Have you tried making an sigma vs Amean plot (by plotSA(fit)) as the User's Guide advises? Have you tried to understand what the plot shows you? The basic idea of filtering is to remove enough lowly expressed probes that the trend in the plot is decreasing.

There are specific filtering strategies available for specific gene expression technology platforms, so it would help if you were more specific about what type of data you are analysing.

ADD COMMENT
0
Entering edit mode

Yes sir, I've already made both a hist(fit$Amean) & a plotSA(fit) as adviced in the limma User's Guide and tried to understand what they showed.

I understand the idea behind filtering by the Amean cutoff. So anyway here's my data and my code :

BiocManager::install(c("breastCancerVDX"))

library(breastCancerVDX)

data(vdx)

vdx

pData(vdx )

erdata <- factor(pData(vdx)$er)

design <- model.matrix(~ erdata )

fit <- lmFit( vdx, design)

hist(fit$Amean)

plotSA(fit)

ADD REPLY

Login before adding your answer.

Traffic: 781 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