DNA sequencing and assembly spans life-altering applications like disease diagnosis to answering questions about our ancestory. Sequencing involves state-of-the-art machines generating nucleic acid sequences (AGCT) from wet samples like blood or salvia, followed by aligning these sequences against known reference sequences. Due to the rapid advancement in sequence generation machines relative to Moore's law, the second step (alignment) has now become the bottleneck. Today's state-of-the-art technology for alignment runs software like BWA-MEM on a cluster of high performance general purpose machines that cannot keep up with the rapid rate of data generated by each new generation of sequencer machines. Recent proposals from academia that claim orders of magnitude alignment speedup come at a cost of significant disruption to the hardware and software currently in use in the industry. In this work, we propose MPU-BWM, a hardware-software solution that achieves orders of magnitude speedup (57 $\times$ over single core x86) on the state-of-the-art BWA-MEM algorithm, with non-intrusive integration to existing processing clusters and with minimal modifications to the BWA-MEM software.