Whole-Genome Shotgun Assembler

File Release Notes and Changelog

Release Name: 4.4

Notes:
Improvements in version 4.4:

    * The runCA executive script is now intelligent enough to pick the correct binary for the host it is running on.  This allows for easy invocation in heterogeneous environments, say, a mix of Solaris and Linux.  Options 'binRoot' and 'grid' have been removed, and option 'pathMap' added.
    * The mer overlapper now fully supports grid operation.  Please see the new section on the mer overlapper in the runCA documentation.
    * The runCA option -p <assembly-prefix-name> is now required.  The previous default was to call the assembly 'asm'.
    * Fasta output is better documented.  Before, we output "all contigs", "scaffolds" and "degenerate scaffolds".  A degenerate scaffold is simply an unplaced contig, and these were reported in the "all contigs" file too.  Now, we output "all placed contigs", "unplaced contigs" and "scaffolds".  See the runCA documentation for more.
    * Replace confusing option gkpBelieveInputStdDev with gkpFixInsertSizes, enabled by default. Gatekeeper (the binary) by default does not fix sizes. runCA by default tells gatekeeper to fix sizes. The required library stdev size range is between 1/10 and 1/3 of the mean.
    * The kmer src is now included. See http://kmer.sf.net.  This enables some new functionality in the wgs-assembler which is off by default, since this is a beta release. It's also known to be broken on 32-bit.

Bugs fixed in version 4.4

    * Non-integer UID support is much better.  Numerous bugs in handling these UIDs were discovered (Thanks, Sven!) and fixed.  Among them -- only 1.6 million 454 reads could be loaded; long UIDs > ~15 letters would cause problems; duplicate UIDs were handled incorrectly.  These errors were usually encountered in the 'gatekeeper' module, sometimes in asmOutputFasta.
    * Consensus was outputting nonsense sequence in the VAR records.



Changes: