stranded vs not stranded Htseq-output differences

Question

[AP]

Hello everyone,

I was not sure if my RNA sequence data were strand specific or not. So while doing the transcript count using Htseq-count, I tried both options and saw there was difference in result. With option stranded=no I found the following result

__no_feature    2924817
__ambiguous     23189
__too_low_aQual 0
__not_aligned   0
__alignment_not_unique  8855974

With option stranded=yes I found the following result

__no_feature    20203107
__ambiguous     163
__too_low_aQual 0
__not_aligned   0
__alignment_not_unique  8855974

This was the result for same sample. As we can see there lots of no feature counts in stranded than in non stranded. Can you please suggest whats going on here and on which I should stick to for further analysis. Will appreciate your suggestions.

Thank you,

Ambika

[Devon Ryan]

Try stranded=reverse, since stranded=yes libraries haven't been made in years.

[AP]

Thank you, will try it and post the results.

[AP]

@Devon Ryan

__no_feature    3188849
__ambiguous     7356
__too_low_aQual 0
__not_aligned   0
__alignment_not_unique  8855974

This is my output from stranded=reverse for same sample. What would you suggest me to go for? I am so confused with these results.

[Devon Ryan]

A dUTP-based library prep was performed, stranded=reverse is the correct setting.

[h.mon]

RSeQC infer_experiment.py may be of help in your situation.

[AP]

Thank you for suggestions.

[h.mon]

You are welcome.

I am so confused with these results.

Did you try it? I am sure it would help clarify your confusion.

Besides, something I should have asked from the beginning: what kit did you (or your sequencing provider used to prepare the sequencing library?

[AP]

H mon, I contacted them and just found out that they did Directional RNAseq. So they made stranded libraries, but in here if we look at my results we find lots of no feature counts on stranded=yes when compared with others. what does this mean? Is it ok to have many no feature counts?

[h.mon]

"Directional RNAseq" is not the name of a library prep kit. But taking into account "Directional RNAseq" and your counts, you have to use stranded=reverse.

If you are so overwhelmed at this step, I strongly suggest you should contact a local bioinformatician for help / teaching / collaboration.

[AP]

Thank you so much for your suggestions.