Entering edit mode

Ellen Sebastian
▴
20

@ellen-sebastian-5895
Last seen 8.4 years ago

Hello everyone,
I am trying to use Limma for microarray background subtraction and
marray
for normalization.
To do this, I need to transfer my data between RGlist (for limma) and
marray objects.
I think that using limma for background subtraction and marray for
normalization is standard, so hopefully someone on here has experience
converting between the two.
I think I'm doing this correctly, but the normalized data I am getting
back
is very different from normalization on the same data I had previously
done
in SMD: the red/green ratios are all very close to 0 (SD < 0.5 per
array).
I know that my previous data-processing scheme was reasonable, so
there
must be something I'm doing wrong with limma/microarray.
I realize this is a very broad question, but does anyone* *see* any
glaring
mistakes that could be corrupting my data and leading to invalid
normalized
red/green ratios?*
In particular, since I've already done background subtraction before
supplying the data to marray, *should I be setting maRb and maGb to
0'*s?
(not doing this introduced a lot of NAs.)
This is my simplified code:
RGraw <- read.maimages(file,"genepix",wt.fun=wtflags(0.1),
verbose=FALSE)
# RGraw: raw red and green fluorescence values from .gpr files
RGsubbed<-backgroundCorrect(RGraw, method = "edwards", verbose =
FALSE)
# RGsubbed : background-subtracted red and green fluorescence
values
rawmarray <-read.GenePix(file)
# retreive all original .gpr data
rawmarray@maRf<-RGsubbed$R
rawmarray@maGf<-RGsubbed$G
#splice in background-subtracted red and green values from limma
to
marray
rawmarray@maGb<-matrix(data = 0, nrow = nrow(rawmarray@maGb),
ncol =
ncol(rawmarray@maGb))
rawmarray@maRb<-matrix(data = 0, nrow = nrow(rawmarray@maRb),
ncol =
ncol(rawmarray@maRb))
#set background values in marray to 0, since I have already
performed
background subtraction in Limma
normMarray<-maNorm(rawmarray, NormMethodNames[n])
#perform normalization using marray
ratios<-maM(normalized)
#retrieve normalized red/green ratios
stdev<-sd(na.omit(ratios))
#these standard deviations are always way lower than what I got
in
SMD, regardless of background subtraction/normalization method
Thanks very much for any help you can offer... I've been banging my
head
against the wall and crawling through browser() for days trying to
figure
this out....
--
Ellen Sebastian
B.S. Candidate, Biomedical Computation
Stanford University, Class of 2015
[[alternative HTML version deleted]]

Possible duplicate of Unexpected output from Limma background subtraction & Normalization