Skip to main content


Fig. 3 | Algorithms for Molecular Biology

Fig. 3

From: Bitpacking techniques for indexing genomes: II. Enhanced suffix arrays

Fig. 3

Timing results for LCP retrieval. Genomes tested are fly (dm5), chicken (gg4), and human (hg19). a Overall time and space usage. b Zoomed image of fastest methods, as bounded by the horizontal dashed line in graph (a). Graphs plot the time in nanoseconds per query as a function of the space required in bytes per genome length. Formats tested are: integer vectors (Int vector); permuted LCP (PLCP) array using a succinct bitarray; direct coding using Elias delta encoding; wavelet trees; differential coding of the PLCP using a BP64-columnar format; direct coding in blocks of 64 (BP64); bytecoding as implemented within SDSL (SDSL byte); and our implementation of bytecoding with or without an exception guide array at an interval of 1024, 256, or 64. Data points for the bytecoding format are joined by lines for each genome, where bytecoding without an exception guide has the slowest time within each group, and a guide interval of 64 has the fastest time. The vertical line in (a) indicates the space usage of permuted LCP, since its running time lies outside the bounds of the graph

Back to article page