v1.3.0.post1

• Make compatible with graph-tool 2.49

v1.3.0

• Fixed a memory leak that caused the homogenize.py script to fail with large datasets
• Fixed a corner case in which some reverse complement sequences were included in the forward component
• Improved multiprocessing load balancing

v1.2.0

• Make NBP source information (source contig, source genome and number of source genomes for the NBP) available to be loaded in Bandage.

v1.1.0.post3

• Make --nice-headers produce contig headers compatible with SqueezeMeta.

v1.1.0.post2

• Fixed an issue in which a cython extension would produce a bugged binary when compiled with conda's gcc

v1.1.0.post1

• Fixed an issue in which the reverse complements were reported instead of the forward components in some cases
• Fixed an issue in which providing the same input genomes in a different order led to slightly different results
• Fixed a rare issue in which certain DBG edges were ignored
• Added the --header-prefix argument to add a custom prefix to sequence names in the FASTA/FASTG outputs
• Added the --nice-headers argument to ensure that sequence names do not contain dashes, to increase compatibility with downstream tools

v1.0.0

• SuperPang now uses 3 times less memory and is around 20% faster. Results should be mostly identical, but bubble detection has improved in some corner cases.
• Added the --lowmem flag to further reduce memory usage by storing data into disk hashes when possible/reasonable. This will significantly reduce speed at the DBG construction stage.
• SuperPang is now published here! Note that the benchmarks shown in the paper come from an earlier version, and as stated above performance has increased significantly.

v0.9.6

• Fix an incompatibility issue with the latest version of graph-tool (2.46)

v0.9.5

• Performance improvements

v0.9.4

• Fixed a bug that happened when the input sequences resulted in a perfectly circular DBG component
• Default completeness has now been increased to 70, which increases mOTUpan's behaviour on viral pangenomes. Note that providing completeness estimates (obtained e.b. with CheckM) is still the recommended way when possible
• The completeness file is now validated before assembly, which saves time in case it is malformed