Hi everyone, I am running nf-core/methylseq v4.2.0 on mouse WGBS data using Bismark/Bowtie2. All 8 samples completed successfully. Sequencing depth is very high, around ~610-783 million total reads per sample, but the Bismark mapping rate is only ~34-37% across samples. After deduplication, ~176-220 million unique reads remained per sample, so usable depth is still high and QC is fairly consistent between groups. I used default trimming and did not include

--accel

or manual hard clipping. M-bias plots show a strong read-end bias, especially at the 5′ end of R2, with elevated CHG/CHH methylation at early positions. I’m planning to rerun with

--accel

. Could lack of Accel/Swift-style hard clipping plausibly explain lower mapping rate, or would you suspect other common causes? Any suggestions on what QC outputs to check before rerunning would be appreciated. @Sateesh_Peri