Should I not have any adapters after trimming?

Question

[bioinfo]

Hello,

I am generating fastq files and I am including the adapter sequences in the samplesheet so they can be trimmed. I used the nextera adapter "CTGTCTCTTATACACATCT". When I run fastQC I see there are between 1-3% of adapters left on the last plot.

Is that a problem? How much adapter content would not affect the results? I am using kallisto afterwards to align the data.

Thank you

Answer 1

[GenoMax]

Having adapters left in the data (if they are actually there) should not affect kallisto results. Most programs will appropriately ignore/soft-clip them during analysis.

Adapters need to be strictly removed if you are doing any de novo work. You don't want any extraneous sequence in your sample in that case.

Answer 2

[swbarnes2]

Kallisto, or aligners like STAR won't care. STAR for sure is happy to soft-clip reads to make the alignment happen. So if there's enough non-adapter there to make the alignment work, it will work. And Kallisto is probably similar...you only want gene counts, so small discrepancies between your read and the reference don't matter if the aligner or pseudo-aligner can assign the read properly to its gene.

Fastqc was made for use with DNASeq. So if you aren't doing DNSeq, a lot of the things it flags are not actually problems, and you can just go without agonizing about how to massage your data into 'passing' fastqc checks.