Hello - I'm new to the bioconductor suite and still don't quite understand how to invoke the exprs class. For example, let's say I'm trying to find genes within the ALL dataset with the most and the least significant effect on age. I've been able to write snippets to find the min/max exprs (expression level) within the set...
> min(exprs(ALL)) [1] 1.984919 > max(exprs(ALL)) [1] 14.12657
...but cannot seem to write a snippet that maps that min/max value to the actual gene?
As for assessing how age effects exprs, this is tricky. Each patient obviously has an age (which is a column) and also an expression level (which as I understand is a function) that maps back to a specific gene. As a start - I thought I'd just plot them but again I'm having trouble with the syntax, I've tried the following from what I've read on the web but to no avail...
> plot(exprs,age(ALL))
and...
> plot(exprs~age(ALL))
Anyone see a glaring problem?
A concept that I cannot quite grasp is from the above command I can see all the column names, there are 21 total. However, when I perform the following command...
> dim(exprs(ALL))
[1] 12625 128
...I can see that I have 12625 genes (rows) and 128 patients (columns). Why don't I see 128 column names with colnames(pData(ALL))
command?
Thanks Martin. I think my issue revolves around writing the proper syntax to call the required pData (data.frame) and exprs (matrix) elements. I think I'll need to develop a loop to help me plot the distribution of p-values for gene effect on age (for all 12625 genes).
Trying to understand 12000 graphs does not sound like the right thing to do at all. Instead you'll want to regress expression on age for each gene, and develop a 'top table' ranking from most to least significant. Good luck!
Hi Martin and other Bioconductors,
I've updated my packages to include limma and developed the following model:
However, I keep getting the following error for the lmFit step (also not showing as a function with help(lmFit).
Why can't I complete this step?
lmFit is in the limma package, so you need to say
or
probably the former is better for interactive use.
Martin, I found this webpage you put together a few years ago (https://www.bioconductor.org/help/course-materials/2016/BiocIntro-May/A3_Statistical_Analysis.html). I'm wondering if its possible to adapt your code so it builds a linear model for age effect on gene expression in ALL. You have the following:
I'd like to switch sex with exprs(ALL).
You're asking a different question, so ask a new top-level post 'building a linear model' or some such, after of course reading the excellent limma manual.