vg

Don't forget to mark the static binary executable:

chmod +x vg

Docker Image: quay.io/vgteam/vg:v1.38.0

Buildable Source Tarball: vg-v1.38.0.tar.gz

Includes source for vg and all submodules. Use this instead of Github's "Source Code" downloads; those will not build as they do not include code for bundled dependencies that the vg build process needs.

This release includes:

• JSON GAM quality string representation of base64-encoded PHRED+0 is now under test
• vg deconstruct now uses OMP tasks instead of nested parallel sections, which caused problems keeping threads caches straight.
• vg giraffe --output-basename is respected again
• Multithreaded GBWT / GBZ construction from GFA in vg gbwt.
• Multithreaded GBZ / GBWTGraph to GFA conversion in vg convert.
• make clean now works on submodules that lack top-level Makefiles
• vg now builds with GCC 11
• vg giraffe --named-coordinates can produce GAM output with positions in the space of the input GFA, for indexes produced from a GFA with vg autoindex.
• CI jobs that use Docker builder deployment now stop early if they don't see it.
• vg autoindex supports --gbwt-buffer-size to increase the buffer size for indexing long haplotypes.
• vg snarls now supports --named-coordinates for reporting snarls and snarl traversals in input GFA coordinates.
• vg mpmap produces more accurate and parsimonious alignments on complex graphs

Updated Submodules

• elfutils
• gbwtgraph
• gssw
• libbdsg
• libhandlegraph
• libvgio
• sdsl-lite
• structures
• vcflib

Make sure to git submodule sync and git submodule update --init --recursive if building from source.

This is a non-versioned release created in order to make a stable DOI for the source code to support reproducibility in the VG pantranscriptomics paper.

Don't forget to mark the static binary executable:

chmod +x vg

Docker Image: quay.io/vgteam/vg:v1.37.0

Buildable Source Tarball: vg-v1.37.0.tar.gz

Includes source for vg and all submodules. Use this instead of Github's "Source Code" downloads; those will not build as they do not include code for bundled dependencies that the vg build process needs.

This release includes:

• vg clip -D fixed to do a better job breaking up giant snarls.
• Option -A added to vg call to call all snarls, including child snarls (analagous to vg deconstruct -a)
• Options -c and -C added to vg call to restrict snarl calling to sites within specified size range
• vg construct -a no longer generates, and other vg tools no longer expect to see, empty paths for some variants
• The segment translation table can be extracted from a GBZ file with vg gbwt.
• vg autoindex no longer calls out-of-scope lambdas captured by reference
• vg call VCF output now more consistent with vg deconstruct: LV and PS tags are added (with -A), snarl names in variant IDs are written the same, and the AT tag is added. -N option added (equivalent to -T in deconstruct) to apply a node translation table.
• vg call GAF output (-T/-G) more informative, with positional information stored in the names and (individual) read support for each allele with the AD optional tag
• .gbz files can now be used anywhere a graph with embedded paths is expected.
• vg autoindex will embed graph paths into Giraffe GBWT and GBZ files.
• vg giraffe will now use the GBZ file for output on a linear reference and no longer needs an xg.
• vg giraffe will prefer to use named paths from an XG, if available, over those in a GBZ.
• Prevents buffer size errors in GBWT creation by vg autoindex

Updated Submodules

• gbwt
• gbwtgraph
• gssw
• libbdsg
• libhandlegraph
• libvgio

Make sure to git submodule sync and git submodule update --init --recursive if building from source.

Don't forget to mark the static binary executable:

chmod +x vg

Docker Image: quay.io/vgteam/vg:v1.36.0

Buildable Source Tarball: vg-v1.36.0.tar.gz

Includes source for vg and all submodules. Use this instead of Github's "Source Code" downloads; those will not build as they do not include code for bundled dependencies that the vg build process needs.

This release includes:

• Builds again on Ubuntu 18.04
• Giraffe records read and pair mapping wall clock times
• Giraffe can no longer output fragment length annotations too large to convert from JSON to GAM
• Giraffe no longer uses duplicate minimizers as often for seeds, potentially increasing mapping speed.
• Giraffe --track-provenance will no longer crash when considering improperly paired mappings
• Surjection can now prune back low-complexity or short anchors to give more-optimal alignments against the target paths (--prune-low-cplx)
• GFA file input is now checked by the parser and not just trusted to match the spec
• Mac CI Homebrew package installation is more robust
• Better handling of reversing reference paths in vg call
• Multiple -P allowed in vg clip
• vg paths messaging cleaned up to acknowledge that vg paths -E -g does indeed need a graph.
• vg giraffe and vg mpmap no longer crash when reads are in lowercase
• vg map -h should now print help again
• CI now supports secure Docker Hub mirrors
• vg call joins deconstruct and surject in transparently supporting reference subpaths
• vg deconstruct changed to write LV and PS tags wrt sites in the VCF (and will not longer reference un-exported sites from the graph)
• vg clip gets -D option for clipping out big deletion edges
• Specified --align-out parameter in vg sim as GAM-format output
• PATH is now quoted when added to

New and Updated Submodules

• gfakluge has been updated.

Make sure to git submodule sync and git submodule update --init --recursive if building from source.

Don't forget to mark the static binary executable:

chmod +x vg

Docker Image: quay.io/vgteam/vg:v1.35.0

Buildable Source Tarball: vg-v1.35.0.tar.gz

Includes source for vg and all submodules. Use this instead of Github's "Source Code" downloads; those will not build as they do not include code for bundled dependencies that the vg build process needs.

This release includes:

• Multipath alignments from vg mpmap are more decomposable in tails near soft clip
• Giraffe does not attempt rescue if it would be too slow.
• vg gamcompare can use distances from a distance index instead of path position annotations.
• Fix bug in vg chunk -T where input path ranges were incorrectly used to find subgraph size.
• Fix GAF parsing bug that caused memory explosion for long paths.
• Fix vg augment -B to not initialize packer object unnecessarily.
• surject joins deconstruct in supporting vg subpath naming, but path lengths must come from --ref-paths in order to be valid in the output header.
• deconstruct AT INFO field fixed to have Number=R instead of A
• vg mpmap has improved sensitivity for detecting spliced alignments and ability to use genome-derived splicing statistics for non-humans
• vg surject -S no longer crashes if it decides that a deleted/spliced segment is too long to realign.
• vg autoindex can handle VCFs that list contigs for which there are no variants
• Giraffe now shuffles tied top read mapping candidates like mpmap and map, to avoid arbitrary biases
• vg sim now produces more realistic RNA-seq fragment length distributions.
• libbdsg dependency should no longer sometimes fail to build on a clean checkout due to missing mio/mmap.hpp\
• Tail alignments can no longer skip read sequence completely when alignment would be too large
• Giraffe progress reports now include fragment length distribution information
• Giraffe can now find .giraffe.gbz GBZ files automatically.
• vg pack gets new -s option to ignore first and last N bases of reads. -s 5 works well in practice for SV genotyping.
• vg call default parameters re-tuned to work a bit better on complex graphs. An extra filter for completely unsupported breakpoints is added to increase precision.
• libvgio now uses a simpler CMake build
• vg call -e now works again
• CI testing now uses Toil 5.4.0
• vg rna now supports using GBWT haploytypes as reference paths (--use-hap-ref option).
• vg rna can now create a spliced pangenome graph and pantranscriptome with chopped nodes (--max-node-length option).
• Spliced pangenome graphs and pantranscriptomes can be sorted and have their ids compacted in vg rna (currently only works for inputs in the PackedGraph format).
• The transcript haplotype info table from vg rna no longer contains the "Reference" column (is part of the "Haplotypes" column now).
• The names for the options --use-all-paths and --add-non-ref-paths in vg rna have been changed (single-letter option the same) and the option --gbwt-id-interval has been removed (not relevant anymore).

New and Updated Submodules

• libbdsg and libvgio have been updated.

Make sure to git submodule sync and git submodule update --init --recursive if building from source.

Don't forget to mark the static binary executable:

chmod +x vg

Docker Image: quay.io/vgteam/vg:v1.34.0

Buildable Source Tarball: vg-v1.34.0.tar.gz

Includes source for vg and all submodules. Use this instead of Github's "Source Code" downloads; those will not build as they do not include code for bundled dependencies that the vg build process needs.

This release includes:

• Support for the GBZ format (GBWT + GBWTGraph):
  • vg autoindex: --workflow giraffe outputs GBZ instead of GBWT and GBWTGraph.
  • vg convert: GBZ -> other graph types.
  • vg gbwt: GBWT + GBWTGraph <-> GBZ.
  • vg giraffe: GBZ as the default input format for GBWT + GBWTGraph.
  • vg index: Graph format for distance index construction.
  • vg minimizer: An alternative to GBWT + graph.
  • vg snarls: Graph format with options that do not need embedded paths.
• vg convert outputs HashGraph instead of VG by default.
• GBWT / GBWTGraph / GCSA2 structures are serialized as bare files for better compatibility with external tools.
• vg convert -w can now handle separator tokens in contig names
• Docker base images are now pulled through Google's cache
• vg giraffe now uses IndexRegistry instead of IndexManager and demands that passed file names already exist
• Wiki now includes a tutorial under test for Giraffe
• vg benchmark evaluates the minimum distance index
• Improved build out of the box on M1 Macs with Homebrew
• vg deconstruct now writes "Allele Traversals" (AT field in INFO) for each site. The traversals map the alleles to paths in the graph.
• IntegratedSnarlFinder properly generates single-node chains for bridge edges in the Cactus graph.
• Allow seeding path traversals from XG path set even when we are using GBWT input for vg deconstruct.
• vg index, vg minimizer, and vg snarls can load GBZ without a distinguishing option
• vg mpmap now uses the RNA mapping preset (--nt-type rna) by default instead of DNA (--nt-type dna)
• Fixed a bug that sometimes caused incomplete multipath alignments from vg mpmap
• IntegratedSnarlFinder properly tracks whether cycles and bridge edge paths have been reached regardless of orientation
• Minimizer indexes no longer incorrectly cache distances for looping chains, fixing a duplicate-alignment, low-MAPQ bug in Giraffe
• DYNAMIC now has hopscotch-map as a sub-submodule
• vg autoindex can create GBZ files for vg giraffe using a GFA with W-lines as input
• The speed of vg mpmap is more robust to low-quality RNA-seq reads
• Multipath alignments can be viewed in dot format with vg view -Kd
• vg clip added to remove alt alleles in bed regions or low-path-depth alleles from the graph

New and Updated Submodules

DYNAMIC, gbwt, gbwtgraph, libvgio, libbdsg, gssw, and sdsl-lite have been updated.

Make sure to git submodule sync and git submodule update --init --recursive if building from source.

Don't forget to mark the static binary executable:

chmod +x vg

Docker Image: quay.io/vgteam/vg:v1.33.0

Buildable Source Tarball: vg-v1.33.0.tar.gz

Includes source for vg and all submodules. Use this instead of Github's "Source Code" downloads; those will not build as they do not include code for bundled dependencies that the vg build process needs.

This release includes:

• vg call -a fixed to not include redundant base at beginning of alleles
• "path end" field corrected in GAF output
• VG now supports ARM Macs.
• vg recalibrate has been removed (deprecated since 1.30.0).
• vg find kmer output also lists samples/phases for the matching haplotypes.
• vg's GAF parser now reads cg cigar strings when no cs cigars found so, for example, vg pack can now properly load a GAF from GraphAligner.
• vg gbwt -R now removes multiple samples correctly.
• vg deconstruct can more easily use reference paths (or subpaths) from the GBWT
• vg mcmc now accepts non-vg::VG graphs as input
• vg deconstruct fixed to not write sites with no alt alleles when genotyping with a gbwt

New and Updated Submodules

The DYNAMIC, gbwt, gbwtgraph, gcsa2, libhandlegraph, libvgio, and vcflibsubmodules have been updated. Thehtslibandtabixppsubmodules have been added. Thevowpal_wabbit` submodule has been removed.

Make sure to git submodule sync and git submodule update --init --recursive if building from source.

Don't forget to mark the static binary executable:

chmod +x vg

Docker Image: quay.io/vgteam/vg:v1.32.0

Buildable Source Tarball: vg-v1.32.0.tar.gz

Includes source for vg and all submodules. Use this instead of Github's "Source Code" downloads; those will not build as they do not include code for bundled dependencies that the vg build process needs.

This release includes:

• CI now tests with sequenceTubeMap
• vg map and vg mpmap now annotate properly-paired read mappings
• vg stats -a reports proper pairing statistics
• Faster graph to GFA conversion with vg convert.
• vg surject now gives a helpful warning if you try to use an unsurjectable GAF, rather than just crashing
• GBWT index can now be used to find alt alleles in vg deconstruct with -g
• Single nodes at the end of bridge paths in the Cactus graph won't show up twice in the snarl decomposition
• Giraffe self-profiling is improved and a standardized speed test script is included
• vg construct can accept multiple --region arguments for different contigs
• Convert GBWTGraph to GFA with vg convert.
• VCF files with * alleles are now accepted by vg construct, and only the particular variant records with * alleles are rejected and warned about
• Uncompressed GFA files can now be directly input (or streamed via stdin) as graphs into vg commands. (previously vg convert -g was required beforehand)
• Karger-stein min cut alogirtm and alt proposal distribution added to vg mcmc genotyper. Scripts for performance analysis are in scripts/MCMC_makefile.sh, scripts/mcmc_dist.sh.
• Count the number of nodes with self-loops with vg stats.
• vg srpe has been removed
• vg mod options --kill-labels and --strong-connect have been removed.
• vg stats -L counts self-loops.
• vg stats -e lists nondeterministic edge sets.
• vg find option --mapping enables finding nodes that map to the selected ids.
• vg is now able to build bundled htslib on Autoconf 2.70+

New and Updated Submodules

The fermi-lite, gbwt, gbwtgraph, 'gcsa2', gfakluge, jemalloc, libvgio, and sdsl-lite submodules have been updated.

Make sure to git submodule sync and git submodule update --init --recursive if building from source. The jemalloc update is also known to not build over top of a previous built version; you may need to rm -Rf deps/jemalloc and git submodule update --init --recursive to fix it in your copy.

Don't forget to mark the static binary executable:

chmod +x vg

Docker Image: quay.io/vgteam/vg:v1.31.0

Buildable Source Tarball: vg-v1.31.0.tar.gz

Includes source for vg and all submodules. Use this instead of Github's "Source Code" downloads; those will not build as they do not include code for bundled dependencies that the vg build process needs.

This release includes:

• Large variants are now always handled as flat, and not aligned to themselves, in vg construct.
• rGFA files (such as those output by minigraph) can now be imported into vg with vg convert -g
• vg autoindex can now construct indexes for the vg mpmap/rpvg transcriptomic pipeline and the vg giraffe tool
• Cigar overlaps replaced with *'s and trailing tabs removed to make GFA output more compliant with 1.0 spec.
• GBWT and GBWTGraph construction from GFA input.
• vg convert can convert GBWTGraph to other graph formats.
• Fixed typo in the IntegratedSnarlFinder test case description
• XG indexes now take somewhat less space (especially those with many paths).
• vg convert can output vg paths (that adhere to certain naming conventions) to GFA W-lines with the -w option.
• Build should fail less on GCC 10.2.1-1.
• vg convert can now output GFA, with selected reference paths written to "rank 0" rGFA tags
• vg combine no longer loads graphs as, and writes graph to, protobuf. Also fixed to join id spaces if necessary.
• vg now uses the vgteam fork of SDSL.
• CI now moved to Gitlab and Github Actions.
• vg autoindex can now simplify plans with joint recipes
• IndexRegistry uses per-instance temp directories
• vg Docker build is now optimized for layer caching
• vg can now build on the arm64 architecture and pass unit (but not CLI) tests
• Updated to latest vgteam version of the DYNAMIC library
• Converting GFA with P-lines and W-lines into GBWTGraph stores the P-lines under sample name _gbwt_ref.
• vg convert from GBWTGraph to another graph type stores the threads for a specified sample (default _gbwt_ref) as paths.

New and Updated Submodules

• DYNAMIC, fermi-lite, GBWT, GBWTGraph, GCSA2, gssw, libbdsg, mmmultimap, sdsl-lite, ssw, and xg have been updated.
• mio and atomic_queue have been added.

Make sure to git submodule sync and git submodule update --init --recursive if building from source.

Don't forget to mark the static binary executable:

chmod +x vg

Docker Image: quay.io/vgteam/vg:v1.30.0

Buildable Source Tarball: vg-v1.30.0.tar.gz

Includes source for vg and all submodules. Use this instead of Github's "Source Code" downloads; those will not build as they do not include code for bundled dependencies that the vg build process needs.

This release includes:

• vg paths now has -C option to list which paths are cyclic and which aren't
• fixed vg view bug which caused outputting to GFA to use too much memory
• Giraffe can now --show-work to explain itself.
• Giraffe funnel plots no longer invent fake candidates
• Giraffe now maps and caps MAPQs properly with syncmers
• vg mpmap produces more complete multipath alignments around indels in low complexity sequences.
• Update libvgio dependency (fixed cmake handlegraph dependency)
• The following subcommands are now deprecated: combine, compare, explode, join, kmers, locify, msga, recalibrate, sift, sort, srpe, translate.
• The following subcommands have been deleted: bugs, crash.
• Crash reports now direct users directly to the new issue URL
• rocksdb is no longer a dependency.
• vg index no longer will produce RocksDB indexes of GAMs or graphs.
• vg find can no longer be used to query from RocksDB indexes.
• vg genotype no longer accepts RocksDB GAM indexes, and instead accepts only GAMs.
• vg gamsort can no longer sort using the RocksDB order.
• vg compare has been removed.
• vg locify has been removed.
• The following subcommands are now deprecated: concat, explode, join, kmers, msga, recalibrate, sift, sort, srpe, translate.
• vg prune is now more memory efficient.
• Giraffe mapping quality cap is now more numerically stable and fails assertions less
• The roc plot script now runs

New and Updated Submodules

A submodule for the wiki has been added in doc/wiki.

Thelibbdsg, libhandlegraph, and libvgio submodules have been updated.

The rocksdb submodule has been removed.

Make sure to git submodule sync and git submodule update --init --recursive if building from source.

Don't forget to mark the static binary executable:

chmod +x vg

Docker Image: quay.io/vgteam/vg:v1.29.0

Buildable Source Tarball: vg-v1.29.0.tar.gz

Includes source for vg and all submodules. Use this instead of Github's "Source Code" downloads; those will not build as they do not include code for bundled dependencies that the vg build process needs.

This release includes:

• XG indexing now uses the temporary directory also for SDSL temp files.
• New GBWT construction interface in vg gbwt.
• vg mpmap now can perform multiple spliced alignments on one read and spliced alignments on multiple mappings.
• Update to latest vcflib ("Configure for default, libcurl-using HTSLib config") to get non-x86 support
• Cross building for other architectures is now easier due to the use of $PKG_CONFIG for specifying pkg-config binary
• SAM/BAM/CRAM TLENs should now match BWA's for identical alignments
• Remove multiple samples with vg gbwt -R.
• Mac builds on Catalina no longer fail with a "missing separator" error.
• CIGAR strings in SAM/BAM/CRAM should now consistently coalesce runs of adjacent insertions and deletions.
• Readme no longer encourages using the unsupported GNU GCC build on Mac.
• Spliced multipath alignments are strand consistent and have a more complete representation of variation near splice sites.
• vg filter can now work on GAMP files with -M, and reads with a given subsequence can be filtered with -a and -A.
• vg mod -n should now work properly when self loops happen at the ends of merged nodes
• CIGAR strings for spliced surjection from GAM files now have correct intron lengths
• CI now tests macOS 10.15
• vg now uses (and needs!) system Boost for Vowpal Wabbit instead of bundling a subset
• Parallel GBWT merging algorithm in vg gbwt.

New and Updated Submodules

The gbwt, mmmultimap, vcflib, vowpal_wabbit, and xg submodules have been updated.

The source repository for the vowpal_wabbit submodule has moved.

The boost-subset submodule has been removed.

Make sure to git submodule sync and git submodule update --init --recursive if building from source.

New dependencies

Instead of shipping a subset of Boost, vg now requires Boost to be installed on the system already in order to build.

Changes in this release:

• vg surject now supports GAF input with -G.
• vg surject can now take GAMP files from vg mpmap as input with -m
• vg surject is now substantially faster
• vg mpmap can now toggle between output formats (including single path and linear formats) with a single option. Some command line arguments have changed semantic meaning.
• vg mpmap's spliced alignment is now significantly more accurate.
• Divided handles with reversing self-edges no longer drop the edge.
• New path cover GBWT defaults: n = 64 for local haplotypes and 16 otherwise.
• Giraffe now properly sets paired end read cross-references
• vg deconstruct includes new option -a to process every snarl (instead of just top-level snarls)
• vg mpmap -m now correctly removes the full length bonus from reported scores
• -lpthread and -lm can be linked dynamically again
• Base quality-adjusted scoring now adjusts the full length bonus.
• Updated with less-patched Raptor dependency
• vg convert calls with -g as the last option will now convert to the correct format, not always Protobuf
• Now vg outputs raw handle graph files readable by libbdsg without the need for vg view -x/--extract-tag.
• vg msga handles terrible alignments better.
• vg giraffe can now output to SAM/BAM/CRAM directly
• Dependency retrieval no longer upgrades already-installed packages
• vg giraffe, vg map, vg mpmap, and vg surject now take a .dict file to order HTSlib sequence headers via --ref-paths.
• GaplessExtender no longer matches non-ACGT characters.
• SAM header lines in dict files are now tested
• vg add now supports overlapping reference paths, as in Cactus graphs

New and Updated Submodules

• libvgio, raptor, dozeu, and vcflib have been updated.
• htslib has been removed. We now use the HTSlib from vcflib.

Don't forget to mark the static binary executable:

chmod +x vg

Docker Image: quay.io/vgteam/vg:v1.27.1

Buildable Source Tarball: vg-v1.27.1.tar.gz

Includes source for vg and all submodules. Use this instead of Github's "Source Code" downloads; those will not build as they do not include code for bundled dependencies that the vg build process needs.

This release includes:

• No longer crashes when converting GFAs that lack a header line
• GAF support in vg paths
• GAF support in vg augment
• vg mpmap can now produce spliced alignments when using the rna preset
• vg gampcompare now handles spliced GAMP files correctly
• Build an r-index with vg gbwt.
• vg snarls -e will output a traversal for every embedded path through each snarl, regardless of snarl complexity
• Refactored vg gbwt.
• Fixed a rare GaplessExtender bug.
• Use the same temp directory for all submodules in vg index.
• Faster parallel GAF input streaming, especially in vg pack.

New and Updated Submodules

The gbwt, gfakluge, and libvgio submodules have been updated.

Make sure to git submodule update --init --recursive if building from source.

Don't forget to mark the static binary executable:

chmod +x vg

Docker Image: quay.io/vgteam/vg:v1.27.0

Buildable Source Tarball: vg-v1.27.0.tar.gz

Includes source for vg and all submodules. Use this instead of Github's "Source Code" downloads; those will not build as they do not include code for bundled dependencies that the vg build process needs.

This release includes:

• mpmap's agglomerate option fixed for paired-end reads

This is a patch release intended to serve as a the version of vg for the paper on the vg giraffe mapper.

Don't forget to mark the static binary executable:

chmod +x vg

Docker Image: quay.io/vgteam/vg:v1.26.1

Buildable Source Tarball: vg-v1.26.1.tar.gz

Includes source for vg and all submodules. Use this instead of Github's "Source Code" downloads; those will not build as they do not include code for bundled dependencies that the vg build process needs.

This release includes:

• Multipath alignments have correct copy and move operators
• vg annotate -m can annotate reads with a path position for each Mapping to a node
• vg annotate now actually respects passed search limit (-l)
• Align promising partial extensions in Giraffe.
• gampcompare now returns the correct distances with the -d option
• vg mpmap -a now agglomerates multipath alignments into a single disconnected alignment (-a no longer controls the maximum alignments through a snarl)
• mpmaps MAPQ now incorporates multiplicities from cluster and alignment hard caps
• vg depth now supports GAF input
• "no sequence" mode using vg view -du where the node sequence is replaced by its size, e.g. to visualize large graphs with dot
• vg sim -E now produces correct positions in single-end simulations
• vg gampcompare output now includes a column to indicate if a read was mapped
• vg surject -S no longer produces occasional malformed BAM records
• Improved heuristic for choosing the extensions to align in Giraffe.
• New option -a for vg call includes 0/0 calls. It can help compare and merge VCFs from multiple samples called against the same graph.
• vg mpmap -a no longer crashes on read pairs that fail to find a consistent pair of mappings
• Update giraffe mapping quality caps

New and Updated Submodules

The raptor submodule has been updated.

Make sure to git submodule update --init --recursive if building from source.

Don't forget to mark the static binary executable:

chmod +x vg

Docker Image: quay.io/vgteam/vg:v1.26.0

Buildable Source Tarball: vg-v1.26.0.tar.gz

Includes source for vg and all submodules. Use this instead of Github's "Source Code" downloads; those will not build as they do not include code for bundled dependencies that the vg build process needs.

This release includes:

• vg gbwt -a adds a path cover of components without haplotypes to an existing GBWT. This resolves the Giraffe issue with a large number of incorrect mapq 60 reads in some graphs with the full GBWT.
• New seeding algorithm targeted at miRNA
• More flexible build system
• Pangenome ontology now has linkZoomLayer
• Synchronized semantics for OVERLAP_THRESHOLD
• Chopping with mod -X changed to run natively on handle graphs
• vg stats -F prints graph format
• Major vg index refactor.
• Always build GBWT from a single graph with a single thread source (VCF, GAM, GAF, paths).
• Index embedded paths as samples instead of contigs with vg index option --paths-as-samples.
• Correct names for long options in vg gbwt.
• Ignore metadata in empty indexes when merging GBWTs.
• add vg stats -D option to print degree distribution
• GAM/GAF I/O code moved into libvgio
• Ability to specify contig ploidy in vg sim
• Alignment with dozeu can now handle large full length bonuses without crashing
• dozeu is now non-x86 compatible.
• Made match pruning in mpmap less aggressive for very short reads
• mpmap can now use the distance index's algorithm for rescue graph extraction
• mpmap now can use the distance index in place of a snarl manager to guide the structure of multipath alignments
• Better vg map error message when xg input not found
• GBWT construction from GAM/GAF is faster and handles duplicate read names correctly.
• vg sim can now output a table of the path positions that reads were simulated from
• Added a subcommand gampcompare to evaluate mapping correctness of multipath alignments
• libvgio now only tries to find Protobuf once
• Memory should no longer blow up in mpmap on splice-abundant graphs
• GaplessExtender returns all non-overlapping full-length extensions.

New and Updated Submodules

The gssw submodule is pointing at a different repository. Use git submodule sync to update the pointer if building from source.

The libvgio, dozeu, gbwt, libbdsg, gssw, structures, and gbwtgraph submodules have been updated.

Make sure to git submodule update --init --recursive if building from source.

Don't forget to mark the static binary executable:

chmod +x vg

Docker Image: quay.io/vgteam/vg:v1.25.0

Buildable Source Tarball: vg-v1.25.0.tar.gz

Includes source for vg and all submodules. Use this instead of Github's "Source Code" downloads; those will not build as they do not include code for bundled dependencies that the vg build process needs.

This release includes:

• New development in vg mcmc
• Graph Alignment Format (GAF) output support for vg map and vg giraffe
• Revisions to distance index format; distance indexes will need to be rebuilt
• STL-based multipath alignment internal format
• Bugfixes to Integrated Snarl Finder
• Improvements to dozeu alignment algorithm
• Bugfixes in banded global aligner algorithm
• CI tests rearranged into batches
• Bugfix for slow clustering in vg giraffe
• Parallel snarl finder in vg deconstruct
• Bugfixes to NetGraph in the absence of unary snarls
• Update to XG encoding
• Transfer mapqs when converting multipath alignments
• Bigger bins for coverage stats in vg call
• Better write parallelism in mpmap
• Case insensitivity for some mapper option values
• Fix build of SNarlManager with latest Protobuf which prohibits inheriting from messages
• Find the missing start indexes on multipath alignments
• New default gbwt id interval value in vg rna
• Renamed vg gaffe to vg giraffe
• Avoid out-of-bounds CNV and inversion breakpoints when generating random test graphs
• Bugfixes to variant caller
• Disqualify ultrabubbles with tips
• Fix vg rna crash when using reverse embedded paths
• Improvements to vg sim read simulation, especially for microRNA
• Don't enter infinite loop in vg sim when simulating from short paths
• Reduce information copying on multipath alignments
• Changed default vg giraffe parameters and added a fast mode
• Fix incorrect assert in vg sim
• Fix presets in vg mpmap
• Made clustering distance limit adjust itself for longer read lengths for vg giraffe
• Fix missing metadata for multipath alignments
• Added vg call option to output traversals to gentoype with rpvg
• Update CI to latest toil-vg
• Better Giraffe rescue
• Ploidy option for vg deconstruct
• Fix annotation boundary bug in vg rna
• Added -T option in vg call to pad traversals
• Added overlap threshold in GaplessExtender to deduplicate nearly identical vg giraffe results
• New seeding algorithm in vg mpmap targeted at miRNA

New and Updated Submodules

The gbwtgraph and xg, submodules have been updated.

Make sure to git submodule update --init --recursive if building from source.

Don't forget to mark the static binary executable:

chmod +x vg

Docker Image: quay.io/vgteam/vg:v1.24.0

Buildable Source Tarball: vg-v1.24.0.tar.gz

Includes source for vg and all submodules. Use this instead of Github's "Source Code" downloads; those will not build as they do not include code for bundled dependencies that the vg build process needs.

This release includes:

• Bugfixes in vg mpmap
• MEM finding heuristic in vg mpmap
• Pointers to Gimbricate and Seqwish when trying to import a non-blunt GFA
• A new IntegratedSnarlFinder that can be faster and use less memory, accessible through vg snarls --algorithm integrated
• Use of dozeu library for vg gaffe rescue
• Faster default mapping parameters in vg rna
• Fix for excessive memory use bug in spliced vg surject
• Fix for an off-by-one error in vg annotate gff parser
• Fix for getting seed distances from distance index
• Additional spliced vg surject bugfixes
• Added option to prespecify topological order for dozeu
• Top level clustering improvements for vg gaffe
• vg add can now work with various graph formats as input
• Made banded alignment traceback not recursive
• Minor optimizations in vg gaffe
• Added GBWT traversals option for vg call
• Shook more bugs out of mpmap presets and alignment
• Added a GBWT traversal finder
• Improved Turtle Pangenome Ontology
• Adjusted FTP URL where CI testing looks for its GIAB test data
• More use of dozeu in vg mpmap
• Support for quality-adjusted alignment in dozeu
• Revised vg mpmap CLI for better usability
• Added a no clustering option to vg mpmap
• Registered magic numbers for libbdsg graph types so they can be loaded without VPKG encapsulation
• Added node qualities to vg pack output format
• Updated internal dozeu and X-drop alignment interface
• Added ability to simulate from a specific sample in vg sim
• Update handle graph API to include node ranks
• Changed XdropAligner usage to be one per thread.
• Removed support for reading GFAs with overlaps: please preprocess with Seqwish to squash the overlaps out
• Cap the number of threads used in the Docker build to a reasonable number for Quay potatoes
• Automask IUPAC codes to Ns in vg construct, because the internal data model can't use them. If you want them in the graph, make them into variants.
• Chunk long paths when extracting paths Protobuf with vg paths
• Bugfixes to the ODGI graph implementation
• Boosted timeout for simulated chr21 CI mapping test
• Tweaked error rates used for vg call
• Substantial bugfixes in the FlowCaller
• Prohibit empty nodes in xg graphs
• Support for ODGI format in vg convert and as an input format more generally
• Added libbdsg, xg, and vg's algorithms and io namespaces to vg Doxygen docs
• Fixed excessive recursion when dealing with snarls
• Added MAPQ support for paired-end vg gaffe.
• Added ability to convert GFA files to other graph formats in vg convert

New and Updated Submodules

The dozeu, gbwt, gbwtgraph, gfakluge, gssw, libbdsg, libhandlegraph, pinchesAndCacti, structures, and xg submodules have been updated.

Make sure to git submodule update --init --recursive if building from source.

Don't forget to mark the static binary executable:

chmod +x vg

Docker Image: quay.io/vgteam/vg:v1.23.0

Buildable Source Tarball: vg-v1.23.0.tar.gz

Includes source for vg and all submodules. Use this instead of Github's "Source Code" downloads; those will not build as they do not include code for bundled dependencies that the vg build process needs.

This release includes:

• vg construct now rejects not-yet-implemented * allele from new VCF spec
• vg CI now runs on Kubernetes
• vg Docker can now be built with Kaniko
• vg gaffe now has MAPQ capping logic to reduce the error rate at MAPQ 60
• Default MAPQ settings in vg mpmap have been changed
• vg CI tests now respect the core limit of the container they are running in, instead of trying to use all the host's cores
• A variety of vg mpmap options were removed. Some short options changed their meanings.
• Mapping real RNA data with vg mpmap is now faster
• vg CI tests now use an updated toil-vg
• vg circularize adds graph edges once again when circularizing paths
• vg gaffe is now faster
• vg mpmap MAPQs are no longer miscomputed in rare cases
• vg gaffe's index finding has been completely redone. It can now create its own indexes or find indexes by prefix. This may lead to it picking up and using indexes that were not passed to it directly.
• Surjection of softclips has been improved.
• vg call now uses a better, flow-based traversal finding algorithm to generate alleles
• vg mpmap uses new algorithms for long-read alignment
• vg CI reports are no longer totally empty for some tests
• The internal XdropAligner is now much simpler
• ProtoHandleGraph is no longer needed and has been removed
• vg concat now works via the HandleGraph API, and has a -p option be path-guided
• HashGraph no longer crashes when incrementing node IDs
• Machinery for haplotype-based rescue in vg gaffe has been added
• XdropAligner now automatically finds the best pinning location

New and Updated Submodules

The gbwtgraph and libbdsg submodules have been updated.

Make sure to git submodule update --init --recursive if building from source.

Don't forget to mark the static binary executable:

chmod +x vg

Docker Image: quay.io/vgteam/vg:v1.22.0

Buildable Source Tarball: vg-v1.22.0.tar.gz

Includes source for vg and all submodules. Use this instead of Github's "Source Code" downloads; those will not build as they do not include code for bundled dependencies that the vg build process needs.

This release includes:

• Support for haplotype transcripts and two training FASTQs in vg sim, as well as bugfixes
• Paired-end support in vg gaffe
• Changes to distance index format to sspport paired-end mapping. All distance indexes will need to be rebuilt.
• Support for all PathHandleGraph implementations in most vg commands. Many commands will now output in formats other than vg Protobuf.
• vg combine command to do what cat does for the Protobuf format when working with graphs in all formats.
• Ability to rename variants' contigs during vg construct
• Updates to many submodules to respect build flags, to make packaging easier
• GFA edge orientations in output are now more canonical
• Removal of vg stats -S and the concept of from/to-siblings
• Addition of new predicates to the RDF ontology
• Addition of man page for vg gaffe
• XG edge vectorization now works correctly, which should make vg pack output and the things that use it more correct.
• Fixed bug with vg chunk -p -C -a when not all paths are wanted in gam chunks
• Added missing critical section in vg chunk
• Fixed haplotype exon break bug in vg rna
• Fixed bugs in new vg chunk gam splitting logic
• Now vg chunk uses all threads by default
• VPKG tag for ODGI files has been fixed. ODGI files will need to be regenerated.
• Now filtering N's in vg augment
• Fixed a bug with vg augment's coverage filter
• Fixed a crash bug in vg surject built with GCC
• Report GQ and a real QUAL in vg call
• Magic number prefixes in SerializableHandleGraph. HashGraph and PackedGraph output will not be readable by older vg versions.
• Fix hash graph id sorting
• For RNA, add option to add splice-junctions from intron bed file
• Speed improvements in vg surject
• Fix vg augment -s option
• Fix rare out-of-bounds index bug in vg mpmap
• Splice support in vg surject
• Add a collective MAPQ option to vg mpmap
• Fix quality string in augmented GAMs
• Ability to simulate RNA reads from haplotype transcripts
• Support for multiple minimizer indexes in vg gaffe
• Addition of an internal dagification overlay HandleGraph
• Disable parallelization of second pass in vg augment
• Do not change names of reference transcripts in vg rna
• Fix an off-by-one error in vg surject
• Add mpmap tests back to CI simulation tests
• Turn on ccache for Travis CI
• More memory efficient vg rna
• MinimizerMapper refactoring
• Update calibration routine for mismapping detection in vg mpmap
• Fully ignore single-node components when doing vg snarls
• Use the HandleGraph API in more places
• Changes to Homebrew usage on Mac Travis CI
• Make filters in vg filter be fully off if not requested

New and Updated Submodules

The gbwt, gbwtgraph, gssw, htslib, libbdsg, libhandlegraph, libvgio, pinchesAndCacti, sha1, sonLib, structures, vcflib, and xg submodules have been updated.

Make sure to git submodule update --init --recursive if building from source.