See #7
Raw data suggests the genome sequence is polymorphic or unreliable at first position of genome (either A or T).
Either we use bowtie2 with global alignment (will probably reduce mapping efficiency) or make some special exception for soft-clips at 5' or 3' end of genome when calculating the consensus.