Can I put single and pair-ended RNAseq data together in DESeq analysis

[ericfit]

I have RNA-seq data from Illumina GA (single-ended)and Hiseq (pair-ended). Just wonder if I could put them together in differential gene expression analysis. I know pair-end is much better than single-end in accuracy. But not sure if it is reasonable to put them together in the analysis. many thanks

[Simon Anders]

No, paired-end is not better in accuracy. At least, for simple differential expression analysis, it does not make much of a difference.

As for your question, the answer is: It depends.

Hence, please always give details on your experimental design when posting a question here,

[ericfit]

Thank you for reply.
I have 10 samples in treatment and 12 samples in untreatment group. In treatment, I have 5 single and 5 pair-ended samples. In untreatment, I have 8 single and 3 pair-ended samples.
Sorry, I am not the person who designed the experiment. Have no idea why they put it into such way. But I have to do the data analysis.

Many thanks.

[Simon Anders]

Well, as long as you have both platforms in both conditions, this is fine. Just treat the library type as a covariate, as in the example in the vignette. (There, we used on purpose a dataset with such a mixture of single-end and paired-end data to demonstrate how to use covariates.)