Getting data with illumina's sequencers, we have sequencing errors in a very large number of reads. Most often they are located at the ends of readings, less often at the beginning. Errors in the last and first nucleotide reads. Why are these errors and not genetic variants? Because these errors are absent in the same positions in intersecting readings. It is not clear where they come from. Whether the sequencer itself creates errors, or whether these are some errors in the preparation of next generation sequencing data. BWA is used as an aligner, strelka2 is used as a variant caller. The error data quality metrics are very good, they cannot be filtered by quality. Help me understand what it is and how to deal with it.