I am working with a previously preprocessed (trimmed, and adapters are removed) miRNA-seq data. On the per base sequence quality plot there are some weird drops in quality at the 2nd, 7th, 14th base. This pattern can be seen in a in about half of the samples.
What can be the cause of it? Should I trim the 5' end?
About the 3' end of the reads: there are just a few reads longer than 30 bp, this yields the low quality and large deviation at the 3' end, I am planning on setting the max read length to 30 or 35 bp.