Persistent Indexing Technology for Large Sequences

There are two aspects to the work being presented here. The first is a novel persistent index structure for genomic data, a prototype of which has been completed. The second, using this index as an example, is a generic index development framework, which is under construction. We propose a variation of the suffix tree, the Top Compressed Suffix Tree, which has been designed to allow the on-disk construction of indexes over multi-gigabyte sequences. This form of the suffix tree extends the work of Hunt et al. [1] by improving the performance of the partitioned construction algorithm when the size of the sequence being indexed is comparable to that of the available main memory, and by providing a compact representation of the index on secondary memory. This work forms part of the GIDOF project—a project to provide a Generic Index Development and Operation Framework. GIDOF addresses the management of performance-critical parameters, automatic parameter exploration and tuning, and the p...

• Type: Conference
• Year: 2003
• Where: BNCOD
• Authors: Robert Japp
• Keywords: BNCOD 2003, BNCOD 2007, Generic Index Development, Index Development Framework, Suffix Tree