**10**wrote:

Background:

Sample Information: 8 groups of cells (cell1,cell2,cell3,cell4,cell5,cell6,cell7,cell8), 4 replicates each group, 32 samples in total.

Comparison we are interested in: cell1-vs-cell2, cell3-vs-cell4,cell5-vs-cell6,cell7-vs-cell8

*Pay attention: sample 1,2 and sample 3,4 in each sample are sequenced in different platform, in an order word, there are batch effect in the samples

I analysed the data twice, with the same operation but with different input. First, I inputed all the data of 32 samples. In the second analysis, I inputed the data only with cell7 and cell8 (8 samples in total). I want to get DEGs of cell7-vs-cell8. In the first time, I got 266 DEGs, while I got none in the second time. It confused me a lot, it shouldn't be the result like that.

I hope someone would offer me some ideas. From all of my heart, sincerely.

How did you evaluate the comparison between cell7 and cell8 in the first analysis? Did you use constrasts?

How have you corrected the batch effect in the second analysis and in the first?

490Thanks for quick response!

The answer to the first question is YES, I made the comparison by using the function

makeContrasts.I didn't remove the batch effect in both two analysis, instead, I specified the batch effect in the linear model, like

design <- model.matrix(~0+group+batch)10The code of my twice analysisThe first analysis:The second analysis:10