Skip to main content
. 2014 Sep 30;2:e606. doi: 10.7717/peerj.606

Figure 3. The PEANUT algorithm.

Figure 3

Read sequences are buffered and a q-group index is created from them on the fly. Filtration (detection of q-gram hits) and validation are performed on the GPU until all reference sequences are processed. The hits are postprocessed and streamed out in SAM format. All layers operate independently in parallel and communicate via queues. Blue layers are I/O bound, green is executed on the GPU, and purple is executed on the CPU. Arrows between the layers denote a data transfer via a queue.