dbsplit(1) divide a Dazzler database into a series of blocks


DBsplit [-a] [-xint] [-sint(200)] path:db|dam


Divide the database path.db or path.dam conceptually into a series of blocks referable to on the command line as path.1, path.2, ... If the -x option is set then all reads less than the given length are ignored, and if the -a option is not set then secondary reads from a given well are also ignored. The remaining reads, constituting what we call the trimmed DB, are split amongst the blocks so that each block is of size -s * 1Mbp except for the last which necessarily contains a smaller residual. The default value for -s is 200Mbp because blocks of this size can be compared by our "overlapper" daligner(1) in roughly 16Gb of memory. The blocks are very space efficient in that their sub-index of the master .idx is computed on the fly when loaded, and the .bps and .qvs files (if a .db) of base pairs and quality values, respectively, is shared with the master DB. Any relevant portions of tracks associated with the DB are also computed on the fly when loading a database block.