I am trying to map Illumina data to rotavirus genome. But I noticed that mapping to different reference genomes results in some variations in the consensus sequence. Any suggestion? Should I discard genome regions with low coverage? If yes, then what is considered low coverage?