Different STRAND in the same analysis

[gevro]

Can samples with different STRAND values be used together in the same analysis?

[vyepez88]

If the values are 'yes' or 'reverse' then yes. If some samples were sequenced not strand-specifically, then no.

[gevro]

The problem is that GTEX control samples are STRAND = no. Does that mean I cannot use them together with my samples that are STRAND = yes?

[gevro]

And if the above is not possible, do you have any suggestions for what to do?

[gevro]

And just to clarify - what I mean is not whether it is recommended to combine stranded with unstranded samples -- I know that is not recommended.
The question is if the pipeline will run at all when combining samples with different STRAND annotations.

[mumichae]

Yes, you can run the pipeline with different strand options, as they are defined for each sample individually. This might lead to batch effects that might or might not be corrected well by the autoencoder. You could either count your samples in a strand unspecific mode or count them strand specifically and evaluate based on the heatmap in the outrider summary, how strong the batch effect is. We would recommend the former, as it minimises known effects.