Skip to content

Latest commit

 

History

History
335 lines (187 loc) · 9.23 KB

NEWS.md

File metadata and controls

335 lines (187 loc) · 9.23 KB
Raw

Release 0.21-r606 (21 April 2024)

Notable changes:

  • New feature: added mgutils-es6.js for converting merged BED to VCF. This replaces the older BED-to-VCF converter. This script requires k8 v1.0.

  • New feature: output a new ds:Z tag. ds is similar to cs but it additionally indicates microhomology around INDELs.

  • New experimental feature: added gafcall.js for calling structural variants. This script will be moved a separate repo in future.

(21 April 2024, r606)

Release 0.20-r559 (20 November 2022)

This release fixes several assertion failures (#78, #79 and #80) caused by recent changes in graph chaining. It also fixes an integer overflow that may happen to chromosomes longer than 256 Mb (#76). Results remain the same.

(20 November 2022, r559)

Release 0.19-r551 (12 June 2022)

This release fixes a segmental fault if minigraph is compiled with certain compiler-libc combinations. This is apparently caused by memcpy(0,0,0). Minigraph is otherwise identical v0.18.

(12 June 2022, r551)

Release 0.18-r538 (9 May 2022)

This release uses heuristics to speed up base alignment in long divergent regions. The heuristics does not guarantee optimal alignment but it reliably produces alignment close to the optimal except in centromeres where the algorithmically optimal alignment may not represent true evolution in biology. The new version is 10-700% faster than v0.17 depending on input data and parameters in use.

(9 May 2022, r538)

Release 0.17-r524 (29 April 2022)

This release adds base alignment to minigraph. It represents the first major improvement to minigraph. Specifically, this release attempts to connect linear chains with the graph wavefront alignemnt algorithm (GWFA) and produces the final alignment with miniwfa under the 2-piece gap penalty. Graph generation also considers base alignment. This gives more accurate graph alignment and generally simpler graph topology. Note that minigraph still focuses on structural variations and does not generate base-level graphs. To endusers, minigraph remains similar feature wise.

Notable changes:

  • New feature: option -c for base alignment and graph generation. In the alignment mode, the option adds the cg:Z CIGAR tag like minimap2. Graph generation still works without -c but applying this option is generally recommended now.

It should be noted that the base alignment is currently slow for species of high diversity. This will be addressed in the next couple of releases.

(29 April 2022, r524)

Release 0.16-r436 (21 February 2022)

Notable changes:

  • Improvement: 2-level chaining. This is a feature backported from minimap2. It speeds up graph generation for human graphs.

  • Improvement: break a chain at poorly aligned regions, another recent minimap2 feature.

  • Added the script for generating figures in the minigraph paper.

(21 February 2022, r436)

Release 0.15-r426 (21 March 2021)

Fixed a bug in bubble identification around inversions. This version should be used together with the latest gfatools for consistency.

(21 March 2021, r426)

Release 0.14-r415 (19 December 2020)

Notable changes:

  • Added the --call option to find the allele/walk in each bubble.

  • Reduced the default minimum variant length (option -L) from 100 to 50 for the consistency with the SV community.

(19 December 2020, r415)

Release 0.13-r397 (3 December 2020)

Notable change:

  • Fixed incorrect anchors in linear chains. In older versions, a linear chain may contain two anchors with identical reference or query coordinates.

(3 December 2020, r397)

Release 0.12-r389 (26 October 2020)

Notable changes:

  • Improve alignments towards ends of graph segments. If there is an SV close to the ends but not at the ends, older versions may produce an excessively large bubble including high-identity matches.

  • Heuristically accelerates alignment in complex subgraphs by skipping many unnecessary sequence-aware graph traversals. This speeds up graph generation for CHM13 by three folds without obviously affecting accuracy.

  • Added option --inv to optionally disable inversions. Graph traversal is hard with inversions.

  • Fixed the bug that prevents large -K.

  • Apply option -K4g to the asm preset.

  • Added option --write-mz to output the positions of minimizer anchors.

(26 October 2020, r389)

Release 0.11-r371 (13 September 2020)

Notable changes:

  • Added option --max-rmq-size to limit the max RMQ size, which is set 100k by default. This heuristic reduces the long running time for aligning long centromeric sequences. The accuracy might be affected in rare cases.

  • Cap the max k-mer occurrence to 250 by default. For maize genomes, the current heuristic may choose an occurrence cutoff larger than 1000. This makes minigraph too slow to be practical.

  • Added option -S to output more detailed information about linear chains.

  • Added option -D to ignore diagonal minimizer anchors. This is useful to mapping a sequence against itself.

(13 September 2020, r371)

Release 0.10-r356 (14 February 2020)

Notable changes:

  • Older releases miss a small fraction of INDELs involving repeats. This release fixes this issue.

  • Added the "stableGaf" command to mgutils.js to convert unstable GAF (e.g. by GraphAligner) to stable GAF.

(14 February 2020, r356)

Release 0.9-r343 (31 December 2019)

Notable changes:

  • RMQ based linear chaining. The chaining accuracy should be higher for large events. The speed remains similar.

  • Use ksw2 to check the sequence divergence of events to be inserted.

  • Treat inversions as special events. Don't insert them as long substitutions.

(31 December 2019, r343)

Release 0.8-r316 (11 December 2019)

This release reduces suboptimal chains caused by the chaining heuristics. It generates slightly simpler human graphs.

(11 December 2019, r316)

Release 0.7-r310 (21 November 2019)

Notable changes:

  • Increased the default maximum INDEL/event length from 10kb to 100kb for assembly mapping and graph generation.

  • Decreased the default minimum INDEL/event length from 250bp to 100bp.

  • Accelerated graph mapping by pre-filtering isolated anchors and disconnected linear chains. This triples the performance when long gaps are desired.

Due to the change of default parameters, this release generates graphs different from the previous versions.

(21 November 2019, r310)

Release 0.6-r302 (17 November 2019)

Notable changes:

  • Assign weight to seeds based on their repetitiveness. This helps chaining in repetitive regions a little bit.

  • For short-read mapping, prefer the reference path if the alternate path is not much better.

Major changes may be coming in the next release.

(17 November 2019, r302)

Release 0.5-r285 (8 September 2019)

Notable changes:

  • Fixed a bug that leads to wrong mapping positions in GAF.

  • Fixed two bugs related to graph chaining.

  • Added option -j to set expected sequence divergence and to adjust other chaining parameters accordingly.

  • Increased the k-mer thresholds for fast divergence estimate. This improves the alignment around low-complexity regions.

  • Tuned the default parameters to add highly divergent events only.

  • Warn about duplicated sequence names in graph construction (#3).

This version generates graphs different from the previous versions. The mapping accuracy is improved due to the bug fixes and parameter tuning.

(8 September 2019, r285)

Release 0.4-r267 (22 August 2019)

Notable changes:

  • Support paired-end mapping for short reads.

  • Remap and calculate coverage (see the new --cov option in the manpage).

  • Fixed multiple edges in the generated graphs. The v0.3 14-genome graph contains one multiple edge.

  • Use dynamic minimizer occurrence cutoff. For human data, the dynamic cutoff is around 137, higher than the default cutoff 100 used in earlier versions. As a result, graph generations will become a little slower.

Due to the last two changes, graphs generated with this version are different from the previous versions.

(22 August 2019, r267)

Release 0.3-r243 (7 August 2019)

This release generates graphs with SR tags on L-lines. The topology of the graph is identical to the one generated with v0.2.

(7 August 2019, r243)

Release 0.2-r235 (19 July 2019)

This release fixes multiple minor bugs. It also considers k-mer matches and improves the accuracy of graph chaining. Nonetheless, the old chaining algorithm, albeit simple, works quite well. The improvement is marginal.

(19 July 2019, r235)

Release 0.1-r191 (6 July 2019)

Initial proof-of-concept release.

(6 July 2019, r191)