@Felix Krueger Thanks, Felix. I will rerun the pipeline with the Accel-style hard clipping. One additional question: can the library-preparation type itself also have a major effect on mapping efficiency? My current run used the default directional Bismark setting, but I am not yet sure whether these libraries were prepared as directional or non-directional, so I need to confirm this with the sequencing core. Just to confirm, if the libraries were actually non-directional but analyzed using the default directional setting, could that also substantially reduce the mapping percentage?