It seems to be established practice to discard short reads (<100nt) which come from 454 sequencing data. When and why should this be done?
When you sequence amplicons (samples amplified by PCR), reads below 100bp tend to be just primers, sometimes with a homopolimeric stretches in between. They can be of a high quality and as such, will make it through simple quality filtering. For a decent overview of errors you can stumble across in 454 see http://bioinformatics.oxfordjournals.org/content/27/13/i304.full
Use of this site constitutes acceptance of our User Agreement and Privacy Policy.
version 2.3.6
Thank you for your answer! The linked article was particularly enlightening.
What if someone was looking for small sequences such as miRNAs in 454 data? Would trying to clean the primers and homopolymers out of the short sequences be worthwhile, or would it be a fool's errand?
Eric, I don't have much experience with sequencing of miRNAs, but it seems perfectly reasonable to clean the primers from reads. You can do all cleaning and filtering in one step using trim.seqs command from Mothur (http://www.mothur.org/wiki/Trim.seqs ).