Fastqc: read end sequence content

Question

[ruth]

I have trimmed adapters (from read start) using cutadapt, however fastqc shows that there is some sequence composition bias in the end of long reads.

My question - how could this happen (probably it's not biological) and should I cut end of such reads? If so, how far should I cut?

Answer 1

[Istvan Albert]

Run fastqc with the -nogroup option to see more precisely what is going on - the image above uses 5 wide bins.

When you cut adapters make sure to also include whatever extra base might be ligated onto the sequence. For example during Illumina sequencing there is an extra A overhang that is not technically part of the adapter yet should be added to the adapter sequence.

[ruth]

Thanks, I will try using -nogroup, sequencing was done using Ion Torrent.

Also, I wanted to ask: is it possible to cut from the end using cutadapt? Or maybe I just can use linux tools to trim last few bases, if so, how to know were is the start/end of read?

[Istvan Albert]

Most tools can be told to cut a fixed number of bases from either end but it would good to identify why such an enrichment would be there and whether that has any significance