Replication Data for: iPuma: High-Performance Sequence Alignment on the Graphcore IPU

This dataset contains the DNA and protein datasets used in benchmarking the IPUMA Smith Waterman alignment implementation for the Graphcore IPU. All dataset files are uploaded as both .fasta and .txt. These contain the same sequences, but are included for compatibility reasons, as some data loaders did not support plain text lines. Unpacking instructions The DNA-large dataset is split into separate parts. After downloading these need to be concatenated again, using a command such as cat .gz.part* > file.gz All .fasta.gz files need to be decompressed before usage. This can be done, eg with unpigz *.gz. For executing the benchmarks .txt files need to be generated from the fasta, that only contain a sequence to be compared in each line. This can be done with the following command: cat file.fasta | awk '(NR+1)%2==1' > file.txt