https://hgpu.org/?p=2314
A parameterisable and scalable Smith-Waterman algorithm implementation on CUDA-compatible GPUs