Search
Question: Using multiple factor analysis in edgeR
0
gravatar for Mahnaz Kiani
12 months ago by
Mahnaz Kiani20
Mahnaz Kiani20 wrote:

Hello,

 

I am trying to analyze my data based on section 3.3 of edgeR and I got stuck in the fist part

design <- model.matrix(~0+Group)
 colnames(design) <- levels(Group)
fit <- glmQLFit(y, design)

 

and getting this error :

Error in terms.formula(object, data = data) : object 'y' not found

 

What 'y" stand for?

 

Thanks,

Mahnaz

ADD COMMENTlink modified 12 months ago • written 12 months ago by Mahnaz Kiani20
1
gravatar for Gordon Smyth
12 months ago by
Gordon Smyth35k
Walter and Eliza Hall Institute of Medical Research, Melbourne, Australia
Gordon Smyth35k wrote:

Well, you're not really at the beginning yet. You first have to read your count data into R before you can do any analysis. 'y' represents the DGEList object containing the read counts. See Chapter 2 of the edgeR User's Guide, especially Sections 2.4 and 2.5.

The title of your question says "multiple factor analysis", but you seem to have only one factor.

ADD COMMENTlink modified 12 months ago • written 12 months ago by Gordon Smyth35k

Thank you for your answer, I do have different factor, I have two genotypes and 3 sampling from a different part of the plant in each genotype. I want to analyze both genotypes together using factor analysis with a genotype: stage interaction term. I did the analysis before using CLC genomics workbench and that gave me the mean counts number from 3 biological reps, therfore my though was that I can use those mean, instead of putting three count data from each biological reads. Can I do that?

 

Thanks,

Mahanz

ADD REPLYlink written 12 months ago by Mahnaz Kiani20
1

No, you should be using the raw counts from the biological replicates. Otherwise there is no way to model the biological variability if you only have a mean count for every genotype-stage combination.

In any case, your description of the experimental design is not clear. Is "stage" the same as the different part of the plant? Or are the different parts of the plant your "biological replicates"? Were all samples in each group taken from the same plant? Please provide a sample table, like that in Section 3.2.4 of the edgeR user's guide.

ADD REPLYlink written 12 months ago by Aaron Lun21k

Thnk you, the table that you asked is as follows:

ADD REPLYlink written 12 months ago by Mahnaz Kiani20
0
gravatar for Mahnaz Kiani
12 months ago by
Mahnaz Kiani20
Mahnaz Kiani20 wrote:
Sample Stem Part Genotype Group Biological rep
1 1 A 1 1
2 1 A 1 2
3 1 A 1 3
4 2 A 2 1
5 2 A 2 2
6 2 A 2 3
7 3 A 3 1
8 3 A 3 2
9 3 A 3 3
10 1 B 4 1
11 1 B 4 2
12 1 B 4 3
13 2 B 5 1
14 2 B 5 2
15 2 B 5 3
16 3 B 6 1
17 3 B 6 2
18 3 B 6 3
ADD COMMENTlink written 12 months ago by Mahnaz Kiani20

Several points:

  1. Add this information as a comment, rather than adding your own answer. You're not answering your own question.
  2. The interpretation of "Biological rep" is still unclear. For example, do samples 1, 4 and 7 come from the same plant? I assume that 1/4/7 and 10/13/16 come from different plants, given that the genotypes are different.
ADD REPLYlink modified 12 months ago • written 12 months ago by Aaron Lun21k

That is right, samples 1/4/7 are from different plants, same genotype.

ADD REPLYlink written 12 months ago by Mahnaz Kiani20

In that case, you can just make a single factor combining "Stem Part" and "Genotype", e.g., A.1, B.1, and so on, and use a one-way layout such as ~0 + Group as you've done in your original post.

ADD REPLYlink written 12 months ago by Aaron Lun21k
Please log in to add an answer.

Help
Access

Use of this site constitutes acceptance of our User Agreement and Privacy Policy.
Powered by Biostar version 2.2.0
Traffic: 381 users visited in the last hour