Exprimental design for DESeq2 or edger

Question

[zizigolu]

Hi

I have 3 different patients : one cancer and two controls

But for cancer I have two time points

I want to know differentially expressed genes beween these time points

I even don't know how to phrase this to search in google

Samples             condition
CS001_time point 1  cancer
CS001_time point 2  cancer
HC002               control
HC003               control

Please give me an idea how I get differentially expressed genes between these time points

Thank you in advance

[Gordon Smyth]

Cross-posted to https://support.bioconductor.org/p/9134936/

[swbarnes2]

With only one replicate for each time point, there's not a lot you can do that's mathematically justifiable.

[zizigolu]

Sorry what if we have two technical replicates for each of these I mean actually I have 2 rounds of sequencing to give greater read depth for each of these

like

Samples                       condition
CS001_time point 1_1    cancer
CS001_time point 1_2    cancer
CS001_time point 2_1    cancer
CS001_time point 2_2    cancer
HC002_1                        control
HC002_2                       control
HC003_1                       control
HC003_2                       control

[rpolicastro]

Technical replicates are usually collapsed, and not treated as biological replicates in the case of differential expression.

[swbarnes2]

You can't sequence the same thing over and over again and pretend you have a handle on biological variation.

[zizigolu]

Sorry does it make any sense to first compare CS001_t1 and CS002 versus HC002 and HC003

then CS001_t2 and CS002 versus HC002 and HC003

and then take uncommon differentially expressed genes as related to time points?

I mean comparing time point 1 plus to another cancer sample versus two controls in one go

and comparing time point 2 plus to another cancer sample versus two controls in next go

And finally comparing the results

[swbarnes2]

You can't play math games to get around a lack of replicates.