andi v0.9.3

This is just a small dotdot-release. Minor changes include:

• Code improvements
• Asymmetric distances are now weighted by coverage
• Long anchors (twice the threshold) are now included in the analyses even if they do not have a partner
• Improved unit tests
• Explicit check for sequence length

Please download the two files libdivsufsort-2.0.2-1.tar.gz and andi-0.9.2.tar.gz and follow their individual installation instructions.

andi v0.9.2

This release fixes a small issue with missing GNU extensions i.e. on OS X. Please download the two files andi-0.9.2.tar.gz and libdivsufsort-2.0.2-1.tar.gz and follow their individual installation instructions.

andi v0.9.1

This is a small release. It adds the manual and fixes some autoconf problems.

Please download the file andi-0.9.1.tar.gz and follow the installation instructions in the readme or in the manual (docs/manual/andi-manual.pdf).

andi v0.9

This release is, again, a significant improvement over the last version. The changes include

• adds a new suffix array construction algorithm
• refactors code to C
• improved autoconf checks
• improved unit tests
• removes RMQ in favour of CLD

andi v0.8.1

The previous distribution did not include unit tests. This is fixed now. Smaller changes include

• improved docs
• improved warnings and errors
• additional unit test for join mode
• monitor code coverage using coveralls.io

andi v0.8

This is the holiday release of andi. It is faster and uses less memory than previous versions, so the numbers in the paper are outdated. The runtime on the Meala 3085 data set is now 6 hours and 21 minutes with just 10.9 GB of memory. That is a 12% speed up and only half the memory. For the smaller datasets the improvements are less noticeable.

Please load the attached andi-0.8.tar.gz and follow the instructions in the Readme.md for building.

The changes include

• Introduction of a novel first variant character (FVC) array, resulting in improved performance
• Fixed bugs on singleton intervals
• Fixed a bug resulting in nan or incomplete comparisons for very closely related sequences.
• Reduced memory usage by freeing the reverse complement of a sequence when it is no longer needed.
• Various micro optimizations.

andi v0.7.3

This is the fastest release of andi, yet. Additional changes (over v0.7.1) include

• Better error handling and error messages
• Better input parsing
• Continuous integration using Travis CI
• Improved unit tests
• Added diagnostics (compile with -DDEBUG and run with -vv to gain additional information)

Please load the tarball andi-0.7.3.tar.gz, unzip it, and follow the build instructions in the readme.

andi v0.7.2

This is a test release.

andi v0.7.1

This version fixes a small error with the build script. As usual, load the andi-0.7.1.tar.gz file, unzip it and follow the build instructions in the readme.

andi v0.7

Warning: Unlike stated in the readme, this release also requires GLIB2 for a successful build. Please load v0.7.1 if this is an issue for you.

This release is significantly better than previous versions of andi. It is up to 4 times faster on small data sets and 7 times for the Maela data set with 3085 genomes. It also features a new low-memory mode, which immensely reduces memory usage.

Please load the andi-0.7.tar.gz file, unpack it and follow the build instructions described in the readme.