I have a RNA-seq data about the plant inoculated with a fungi pathogen.Three biological replicates of the experiment were conducted at separate times and using independently grown plants and fungi. There are one pair of near-isogenic line, two fungi races and a series of sampling times post inoculation.

The experimental design as follow, can anyone hlep me about how to find the differenttial expression genes with edgeR.

The object is to clear which genes confer the resistance of the plant and which genes confer the virulence of the fungi.

Comprehensive understand the interaction between plant and fungi.

The simplest approach here seems to be a one-way layout for the biological terms, with an additive batch effect:

grouping <- factor(paste(plant, fungi, time, sep="."))
batch <- factor(batch)
design <- model.matrix(~ 0 + grouping + batch)

This avoids making any assumptions about the (lack of) interactions between the interesting experimental factors. Otherwise, if you used an additive approach for all of these factors, you would be assuming that, e.g., the effect of time would be the same in different plants or for different fungi, which may not be true.

Anyway, with the design I've described above, it's fairly easy to do comparisons between specific groups. If you want to identify DE genes between, say, NIL-R inoculated with A at 0h with NIL-R inoculated with B at 0h, you could just do something like:

con <- makeContrasts(groupingNILR.B.0h - groupingNILR.A.0h, levels=design)

With this strategy, you can compare any two (or more) groups of interest. For example, you could identify differences in the inoculation effect of A at 18 hours against that of B in NIL-R plants by doing:

con <- makeContrasts((groupingNILR.B.18h - groupingNILR.B.0h) - 
                     (groupingNILR.A.18h - groupingNILR.A.0h), levels=design)

This will compute the log-fold change between 18 and 0 hours after A inoculation, and compare it to the corresponding value after B inoculation. In both cases, note that I've assumed that you've renamed the levels of plants to get rid of the dash, because makeContrasts doesn't like that in the column names of design.

table move to above

Could you kindhearted to help me again? 
In my data, one biological repeat (CYR32_1) in one group is obvious deviation from the other two biological repeats(CYR32_2 and CYR32_3) in the plotMDS with or without the method BCV.

So I just removed this sample, while when the design (= model.matrix(~ 0 + grouping + batch)) was used, it seemed among all of the data only the batch2 and batch3 were used.
design
    groupingNILR.B.0h    groupingNILR.A.0h    groupingNILR.B.18h    groupingNILR.B.24h    groupingNILR.S.18h    groupingNILR.S.24h    groupingNILS.B.18h    groupingNILS.B.24h    batchb    batchc

the levels of con：groupingNILR.B.0h    groupingNILR.A.0h    groupingNILR.B.18h    groupingNILR.B.24h    groupingNILR.S.18h    groupingNILR.S.24h    groupingNILS.B.18h    groupingNILS.B.24h    batchb    batchc

After the statistical analysis with glmFit and glmLTR, there are 11994 significantly differentially expressed genes with the FDR<0.05.

I tried to omit the batch effect to use the design3 (= model.matrix(~ 0 + grouping)). The levels of con are groupingMK.26.usp, groupingMK.32.usp, groupingNR.26.18h, groupingNR.26.24h, groupingNR.32.18h, groupingNR.32.24h, groupingNS.32.18h, groupingNS.32.24h. After the statistical analysis with glmFit and glmLTR, there are only 10398 significantly differentially expressed genes (DEG) with the FDR<0.05. compare with the result of “design”, 8722 was identified by both “design” and “design3”. In addition, the FDR and P-value of the same gene is different.

So my questions are:
1. When one of the biological repeat should be removed, how to design the model?
2. In my data, which result seem to be more receivable, the “design” or “design3”, or some others?
3. Does the batch effects have a great effect of the DEG.
4. During the estimation of dispersion, there are three approaches:
d = estimateGLMCommonDisp(d,design)
d = estimateGLMTrendedDisp(d,design)
d = estimateGLMTagwiseDisp(d,design)
After each method was used, the last logFC of each gene will have a minor alteration while the FDR and p-value will have a great change, in some paper only the first and the third method were used, so I’m not so clear whether all of these three approaches should be used together, for in the edgeR users guide just described “The tagwise dispersion approach is strongly recommended in multi-factor experiment cases.”