Hi, I'm working with BGI DNBseq WGBS for the first time. I ran the Bismark workflow on 5M reads of one sample to check M-bias and adapter content. The M-bias plot shows 20bp of bias in the beginning of R2 and 2bp at the end of the read. 1. Is it fair to assume that the 20bp bias is from end repairing like the classic example of BS-seq here?If so, it is better to ignore those bases during meth calling but leave them in rather them trimming them for better alignment sake?