Hi, a few of the people in our group are having the same problem as when we come to the intersect of significant genes called by EdgeR and Voom (but not DESeq) we have a number considerably greater than 2 (I have 476). The TAs suggested posting this to see if there is something missing in the seminar for our assignment.

for function prepareData, it uses melt, but reshape2 is not loaded with the function. so an error message says 'can not find function melt' will appear when call prepareData later.
line#70-95
line#131-135
(sry I don't know if there is a quick way to create link to which line in script

Mirror what I just did in the devel repo, in preparation for releasing hw01 solution and hw02 assignment

Hi All,

I got stucked at the last part of DESeq section. I followed everything in the seminar (I actually tried downloading the R markdown file and run from there) and when I run plotMA(results) (the last line in DESeq Section), I got this error message:

Error in MA[, array] - x : non-numeric argument to binary operator
In addition: There were 50 or more warnings (use warnings() to see the first 50)

I am not sure what's causing it.

Thanks,
Mang

Figure out and address why it does not fully compile.

See more detailed issue on the private devel-repo.

@sjackman

Note I have opened this issue on the public repo.

I have centralized all instructions about submitting homework, including lots of advice about how to use GitHub.

Do you want to add anything about how to handle, e.g. the issues TAs open on the homework? Answer the recurring question about what people are supposed to do in response. Also say what you want re: closing all HW1 issues prior to submitting HW2?

Rendered version here:
http://www.ugrad.stat.ubc.ca/~stat540/homework/hw_submission-instructions.html

Source here:
https://github.com/jennybc/stat540_2014/blob/master/homework/hw_submission-instructions.rmd

Now that I look at it … I can make that plain Markdown instead of R Markdown which will work nicer on GitHub…..

Q1c asked us to remove probes without matching gene names and to use limma to find hits. The answer to this, which also carries over into Q3, depends on when you remove the probes. There are two possible ways to implement this:

• option 1: from the set of ~10k probes, remove ~5k that have no matching gene, and then do limma on the remaining 5k at FDR 1e-5 (this results in 932 genes)
• option 2: use limma on the full set of ~10k genes, then remove rows without a matching gene (results in 725)

It seems like most students+staff agree that option 2 is more appropriate, but perhaps the question could be rephrased to make that more clear.

R's practice of using row names doesn't play well with GitHub's automatic rendering of tab-delimited, .tsv files. GitHub thinks the first data row has 1 too many fields.

If I want the nice .tsv features to work, I'll have to abandon row names, which I usually do anyway, but had used in this particular example.

Files affected:

• GSE4051_MINI.tsv
• GSE4051_design.tsv
• GSE4051_data.tsv (too large to display anyway, though)

and obviously all the scripts that write and read them!