Output from Trimmomatic paired end data- use the paired survived output?

Question

[datanerd]

Hi all,

I have RNA seq data downloaded from SRA database. I did trim them using the default parameters in Trimmomatic. I get 4 outputs- ones that survived both and others that survived one (either forward or reverse) but not both. Should I use the data that survived both for alignment? Or am I throwing away too many reads?

Question 2-

Could anyone explain what this could mean? the drop in the quality at position 35?

Question 3- How much does deduplication matters and what can we do about it?

Thanks so much!

Mamta

Answer 1

[Devon Ryan]

1. I typically ignore orphans unless a high percentage of reads would otherwise be lost. In particular, note that some tools (namely htseq-count) can't deal with mixed paired and single-end alignment files.
2. There are many reasons for a drop in quality, not all of which would be apparent to an end user. Ask whomever did the sequencing, since they can look at the whole run and see if there were system-wide effects at that point (e.g., a bubble in the lane(s) with your samples).
3. Don't mark/remove duplicates with RNAseq data. There are many posts on this so I won't write the reasons for the hundredth time.

[datanerd]

Thanks Devon! I will proceed with your suggestions. I did read about duplicates but was not sure how much is tolerable.

However since I cannot do anything about the drop in quality, I will just use it for now.

[datanerd]

Hi Devon,

One more quick question-

I see some K mer content bias. Although there are no overrepresented sequences. This shows up at different position in Forward and reverse reads. Could this be a problem if ignored?

Thanks so much in advance!

Mamta

[Devon Ryan]

That's also common with RNAseq data and generally not something to concern yourself with.

[datanerd]

Great!Thanks, now I will be more comfortable moving forward.