1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
|
== Trunk ==
* Changing distance metric to enum in Rabema.
* Lowering memory usage of rabema_build_gold_standard by not storing whole GSI in memory but dumping directly.
* GSI can now be written and read from compressed file.
* A detailed and updated manual is now available.
* Using new command line parser yields better documentation.
* WIT is replaced by GSI file format (gold standard intervals).
* Rabema can handle BAM now as well as SAM.
* Rabema has been split into two programs `rabema_build_gold_standard` and `rabema_evaluate`.
* Sort order of SAM files is now checked on the fly and not at the beginning any more.
* Oracle also works in Rabema 1.1.
* Major restructuring of the code, removing unused parts, debug comments etc.
* Reference sequences are now loaded using an FAI index on demand.
This reduces the memory requirement for reference sequences to storing the longest contig two times (forward and RC).
* Improved the output of Rabema to look nicer, progress displays are more exact.
* Adjusting Rabema to use `bam_io` module and stream through the SAM file.
The only thing to keep resident in memory is a few integers for each read (not alignment) and a mapping of read names to ids.
This greatly reduces the memory consumption, e.g. from X GB to Y GB on the example from the Rabema homepage.
* Added `prepare_sam` tool to prepare the updated RazerS 2 output to be used with new Rabema versions.
The updated fragment store does not output read sequences for secondary alignments.
However, this requires a global view and the read sequences to be always in memory.
The tool accepts a SAM file sorted by query name, copies over sequences from primary alignments to secondary alignments (it knows about paired-end reads).
== Release 1.0 ==
Initial realease for the published paper.
|
