Hello everyone,

When a create a generic tabular in a two-channel microarray, I am asked to choose the foreground and background channels and also a "Flags" column which is optional. I figured it was the way I filter out the bad spots, but nothing happens wether I use or not use that box. In my case, I load GenPix files that have a "Flags" column, so I select that column, but nothing changes. Since limma filters out spots with value 0, I have also tried to replace all the GenPix "bad" or "not found" spots by assigning them the value 0, but once again, nothing changes. Any suggestions?

When I filter spots out of the analysis, does that mean that the expression level of those genes will be ignored? I am particularly interested in removing specific genes from the analysis.

Thanks for your help.

Cheers,

Mauro

Hi Mauro,

your assumption that the flags box was supposed to be used for optionally filtering out spots is right - unfortunately,
due to lack of demand (you are actually the first person to ask for this feature), we haven't implemented the functionality
yet. However, since there is a demand now, I will incorporate the filtering option in the next release of Robin. I 'm afraid
that does not help you right now, though - so until the next release is ready you could still use Robin to do the analysis
without using the flags column and then modify the R code that Robin generates to include the flags. I'll be happy to help
you with this (in the end it's my mistake that the flags box doesn't do anything yet ) - so please don't hesitate to ask!

cheers,
Marc

Thanks for answering Marc. I was going to do what you suggested anyway. I'll let you know how it goes.

Cheers,
Mauro

Hi Marc,

This is what I have done to filter out bad spots: I have added the lines in bold in the R files generated by Robin. Could you please tell me if it is ok?
Thanks!

Cheers,
Mauro


myfun <- function(x) as.numeric(x$Flags > -50.5)

## not generic type - we should be able to read most information from the data files
if (source != "generic") {

RG <- read.maimages(files=targets$FileName, source=source, quote="\"'", columns=columns, wt.fun=myfun )

} else {
## generic type - we need detailed info about the input data

anno <- list("ID")
RG <- read.maimages(files=targets$FileName,
source="generic",
columns=columns,
annotation=anno,
quote="\"'", wt.fun=myfun)
RG$printer <- printLayout
}

Hi Mauro,

yes, that should do the filtering (exclude values below -50.5).

cheers,
Marc