GPU Base Calling for DNA Strand Sequencing

摘要

Emerging third-generation DNA strand-sequencers are achieving unprecedented levels of sensor speed density in a portable platform. A substantial computing effort is need to process the data from these machines in real-time. Among the processing challenges is base calling, a sequence labelling step that associates raw measurements with DNA bases. We have implemented a serial and a GPU based parallel version of the hidden Markov model based Base calling. The throughput of Parallel Base calling is 2.85× faster than the existing tool.