For this step:

dds < - DESeqDataSetFromMatrix（countData = counts，
                                  colData = coldata，
                                  design = ~condition）

I read a lot of examples on the website, but these examples seems for majority cases, here are three parts I couldn't make sure:

1. countData = counts
   Generally, part of counts matrix like the following, conl1 conl2 are two samples,

                 conl1 conl2 rep1 rep2    
   ENSMUSGid1     0     0    3    4    
   ENSMUSGid2     0     1    0    0    
   ENSMUSGid3     0     1   10   12
   

   but for my data, even as for conl1, here are three experimental replications, so, it means, for colnames of my raw countData:
   conl1_1, conl1_2, conl1_3, conl2_1, conl2_2.... I plan just seem all of them(conl1_1, conl1_2, conl1_3) as conl. Is it okay?

2. colData = coldata
   Can this coldata contains all potential variables I want to study, including cell type column, batch column, gender column, age column? And let design = ~condition1 + condition2 +condition3...(Order according to proportion from PCA)? Or study each variable one by one?

3. design = ~condition
   Actually my colnames of raw countData also contains batch information, generally, people use design = ~condition, can I use design = ~batch + condition (I gave one column of batch for colData = coldata)? Or run a design = ~condition DEseq2 result then design = ~batch DEseq2 result? Or I should check batch effects exist firstly, then corrected, then do design = ~condition?

Thanks for your attention in advance, sorry for so many confused... Any ideas of each part will be very grateful!!!