Movatterモバイル変換


[0]ホーム

URL:


Skip to content

Navigation Menu

Search code, repositories, users, issues, pull requests...

Provide feedback

We read every piece of feedback, and take your input very seriously.

Saved searches

Use saved searches to filter your results more quickly

Sign up

C++ library and cmdline tools for parsing and manipulating VCF files with python and zig bindings

License

NotificationsYou must be signed in to change notification settings

vcflib/vcflib

Repository files navigation

A C++ library for parsing and manipulating VCF files.

Github-CIAnacondaBadgeDLBrewBadgeGuixBadgeDebianBadgeC++0xChat on Matrix

Vcflib and related tools are the workhorses in bioinformatics for processing the VCF variant calling format. See

A spectrum of free software tools for processing the VCF variant call format: vcflib, bio-vcf, cyvcf2, hts-nim and slivar.Garrison E, Kronenberg ZN, Dawson ET, Pedersen BS, Prins P (2022), PLoS Comput Biol 18(5): e1009123.https://doi.org/10.1371/journal.pcbi.1009123

(bibtex reference at the end of this page)

news

Humpty Logo - CC0 license

November 2024: a fresh release with, mostly, pangenome related fixes.

January 2023: this is vcflib's firstHumpty Dumpty release:vcfcreatemulti is the natural companion tovcfwave.Often variant callers are not perfect.vcfwave with its companion toolvcfcreatemulti can take an existing VCF file that contains multiple complex overlapping and even nested alleles and, unlike Humpty Dumpty, take them apart and put them together again.Thereby, hopefully, creating sane VCF output that is useful for analysis and getting rid of false positives.

We created these tools by including the state-of-the-artbiWFA wavefront aligner.The tools are particularly useful for the output from structural variation callers and pangenome genotypers, such as used by the Human Pangenome Reference Consortium (HPRC) because of overlapping ALT segments.

See alsoRELEASE_NOTES.md

May 2022: thevcflib paper has been published on PLoS Computational Biology!

April 2022: vcflib has just gone pangenome!

By introducing the wavefront algorithm we can now realign long sequences and reduce call complexity (and FPs!) introduced by pangenome variant callers using the newvcfwave tool.

See alsoRELEASE_NOTES.md

overview

TheVariant Call Format(VCF)is a flat-file, tab-delimited textual format thatdescribes reference-indexed variations between individuals. VCFprovides a common interchange format for the description of variationin individuals and populations of samples, and has become thede facto standard reporting format for a wide array of genomicvariant detectors.

vcflib provides methods to manipulate and interpret sequence variationdescribed by VCF. It is both:

  • an API for parsing and operating on records of genomic variation asit can be described by the VCF format
  • a collection of command-line utilities for executing complexmanipulations on VCF files

vclib is both a library (with an API) and a collection of usefultools. The API provides a quick and extremely permissive method toread and write VCF files. Extensions and applications of the libraryprovided in the included utilities (*.cpp) comprise the vast bulk ofthe library's utility.

We have also added infrastructure to write Python bindings. Seebelow.


Short index:


INSTALL

For latest updates seeRELEASE NOTES.

Conda installs in user land without root access

conda install -c bioconda vcflib

Homebrew installs on Linux and Mac OSX

brew install brewsci/bio/vcflib

For Debian and Ubuntu

apt-get install libvcflib-tools libvcflib-dev

We develop against guix and vcflib is packaged as

guix package -i vcflib

See also the Guix shell below.

USAGE

Users are encouraged to drive the utilities in the library in astreaming fashion, using Unix pipes to fully utilize resources onmulti-core systems. Piping provides a convenient method to interfacewith other libraries (vcf-tools, BedTools, GATK, htslib,bio-vcf, bcftools,freebayes) which interface via VCFfiles, allowing the composition of an immense variety of processingfunctions. Examples can be found in the scripts,e.g../scripts.

TOOLS

filter

filter commanddescription
vcfuniqList unique genotypes. Similar to GNU uniq, but aimed at VCF records.vcfuniq removes records which have the same position, ref, and alt as the previous record on a sorted VCF file. Note that it does not adjust/combine genotypes in the output, but simply takes the first record. See also vcfcreatemulti for combining records.
vcfuniqallelesList unique alleles For each record, remove any duplicate alternate alleles that may have resulted from merging separate VCF files.
vcffilterVCF filter the specified vcf file using the set of filters

metrics

metrics commanddescription
vcfhethomratioGenerates the het/hom ratio for each individual in the file
vcfhetcountCalculate the heterozygosity rate: count the number of alternate alleles in heterozygous genotypes in all records in the vcf file
vcfdistanceAdds a tag to each variant record which indicates the distance to the nearest variant. (defaults to BasesToClosestVariant if no custom tag name is given.
vcfentropyAnnotate VCF records with the Shannon entropy of flanking sequence. Anotates the output VCF file with, for each record, EntropyLeft, EntropyRight, EntropyCenter, which are the entropies of the sequence of the given window size to the left, right, and center of the record. Also adds EntropyRef and EntropyAlt for each alt.
vcfcheckValidate integrity and identity of the VCF by verifying that the VCF record's REF matches a given reference file.

phenotype

phenotype commanddescription
permuteGPAT++permuteGPAT++ is a method for adding empirical p-values to a GPAT++ score.

genotype

genotype commanddescription
hapLrtHapLRT is a likelihood ratio test for haplotype lengths. The lengths are modeled with an exponential distribution. The sign denotes if the target has longer haplotypes (1) or the background (-1).
normalize-iHSnormalizes iHS or XP-EHH scores.
abba-babaabba-baba calculates the tree pattern for four indviduals. This tool assumes reference is ancestral and ignores nonabba-baba sites. The output is a boolian value: 1 = true , 0 = false for abba and baba. the tree argument should be specified from the most basal taxa to the most derived.

transformation

transformation commanddescription
vcfintersectVCF set analysis
vcfleftalignLeft-align indels and complex variants in the input using a pairwise ref/alt alignment followed by a heuristic, iterative left realignment process that shifts indel representations to their absolute leftmost (5') extent.
vcfglxgtSet genotypes using the maximum genotype likelihood for each sample.
vcfkeepsamplesoutputs each record in the vcf file, removing samples not listed on the command line
vcfallelicprimitivesNote that this tool is considered legacy and will emit a warning! Usevcfwave instead.
vcfcommonsamplesGenerates each record in the first file, removing samples not present in the second
vcfcreatemultiGo through sorted VCF and if overlapping alleles are represented across multiple records, merge them into a single record. Currently only for indels.
vcfldCompute LD
vcfsamplenamesList sample names
vcfsample2infoTake annotations given in the per-sample fields and add the mean, median, min, or max to the site-level INFO.
vcflengthAdd length info field
vcfstreamsortSorts the input (either stdin or file) using a streaming sort algorithm. Guarantees that the positional order is correct provided out-of-order variants are no more than 100 positions in the VCF file apart.
dumpContigsFromHeaderDump contigs from header
vcfgeno2haploConvert genotype-based phased alleles within --window-size into haplotype alleles. Will break haplotype construction when encountering non-phased genotypes on input.
vcfinfo2qualSets QUAL from info field tag keyed by [key]. The VCF file may be omitted and read from stdin. The average of the field is used if it contains multiple values.
vcfevenregionsGenerates a list of regions, e.g. chr20:10..30 using the variant density information provided in the VCF file to ensure that the regions have even numbers of variants. This can be use to reduce the variance in runtime when dividing variant detection or genotyping by genomic coordinates.
vcfbreakmultiIf multiple alleles are specified in a single record, break the record into multiple lines, preserving allele-specific INFO fields.
vcfwaveRealign reference and alternate alleles with WFA, parsing out the primitive alleles into multiple VCF records. New records have IDs that reference the source record ID. Genotypes are handled. Deletions generate haploid/missing genotypes at overlapping sites.
vcfafpathDisplay genotype paths
vcfremoveaberrantgenotypesstrips samples which are homozygous but have observations implying heterozygosity. Remove samples for which the reported genotype (GT) and observation counts disagree (AO, RO).
vcf2tsvConverts VCF to per-allelle or per-genotype tab-delimited format, using null string to replace empty values in the table. Specifying -g will output one line per sample with genotype information. When there is more than one alt allele there will be multiple rows, one for each allele and, the info will match the 'A' index
vcfqual2infoPuts QUAL into an info field tag keyed by [key].
vcfnumaltoutputs a VCF stream where NUMALT has been generated for each record using sample genotypes
vcfprimersFor each VCF record, extract the flanking sequences, and write them to stdout as FASTA records suitable for alignment.
vcf2fastaGenerates sample_seq:N.fa for each sample, reference sequence, and chromosomal copy N in [0,1... ploidy]. Each sequence in the fasta file is named using the same pattern used for the file name, allowing them to be combined.
vcffixupGenerates a VCF stream where AC and NS have been generated for each record using sample genotypes
vcfkeepinfoTo decrease file size remove INFO fields not listed on the command line
vcfsamplediffEstablish putative somatic variants using reported differences between germline and somatic samples. Tags each record where the listed sample genotypes differ with . The first sample is assumed to be germline, the second somatic. Each record is tagged with ={germline,somatic,loh} to specify the type of variant given the genotype difference between the two samples.
vcfflattenRemoves multi-allelic sites by picking the most common alternate. Requires allele frequency specification 'AF' and use of 'G' and 'A' to specify the fields which vary according to the Allele or Genotype. VCF file may be specified on the command line or piped as stdin.
vcfglboundAdjust GLs so that the maximum GL is 0 by dividing all GLs for each sample by the max.
vcfechoEcho VCF to stdout (simple demo)
vcfgeno2allelesmodifies the genotypes field to provide the literal alleles rather than indexes
vcf2dagModify VCF to be able to build a directed acyclic graph (DAG)
vcfannotategenotypesExamine genotype correspondence. Annotate genotypes in the first file with genotypes in the second adding the genotype as another flag to each sample filed in the first file. annotation-tag is the name of the sample flag which is added to store the annotation. also adds a 'has_variant' flag for sites where the second file has a variant.
vcfremapFor each alternate allele, attempt to realign against the reference with lowered gap open penalty. If realignment is possible, adjust the cigar and reference/alternate alleles. Observe how different alignment parameters, including context and entropy-dependent ones, influence variant classification and interpretation.
vcfremovesamplesoutputs each record in the vcf file, removing samples listed on the command line
vcfcleancomplexRemoves reference-matching sequence from complex alleles and adjusts records to reflect positional change.
vcfindexAdds an index number to the INFO field (id=position)
vcfannotateIntersect the records in the VCF file with targets provided in a BED file. Intersections are done on the reference sequences in the VCF file. If no VCF filename is specified on the command line (last argument) the VCF read from stdin.
smoothersmoothes is a method for window smoothing many of the GPAT++ formats.
vcfkeepgenoReduce file size by removing FORMAT fields not listed on the command line from sample specifications in the output
vcfcombineCombine VCF files positionally, combining samples when sites and alleles are identical. Any number of VCF files may be combined. The INFO field and other columns are taken from one of the files which are combined when records in multiple files match. Alleles must have identical ordering to be combined into one record. If they do not, multiple records will be emitted.
vcfcatConcatenates VCF files
vcfinfosummarizeTake annotations given in the per-sample fields and add the mean, median, min, or max to the site-level INFO.
vcfclassifyCreates a new VCF where each variant is tagged by allele class: snp, ts/tv, indel, mnp
vcfoverlayOverlay records in the input vcf files with order as precedence.
vcfaddinfoAdds info fields from the second file which are not present in the first vcf file.
vcfnullgenofieldsMakes the FORMAT for each variant line the same (uses all the FORMAT fields described in the header). Fills out per-sample fields to match FORMAT. Expands GT values of '.' with number of alleles based on ploidy (eg: './doc/.' for dipolid).
vcfgenosamplenamesGet samplenames

statistics

statistics commanddescription
permuteSmoothpermuteSmooth is a method for adding empirical p-values smoothed wcFst scores.
plotHapsplotHaps provides the formatted output that can be used with 'bin/plotHaplotypes.R'.
pVstpVst calculates vst, a measure of CNV stratification.
wcFstwcFst is Weir & Cockerham's Fst for two populations. Negative values are VALID, they are sites which can be treated as zero Fst. For more information see Evolution, Vol. 38 N. 6 Nov 1984. SpecificallywcFst uses equations 1,2,3,4.
popStatsGeneral population genetic statistics for each SNP
vcfsitesummarizeSummarize by site
vcfrandomsampleRandomly sample sites from an input VCF file, which may be provided as stdin. Scale the sampling probability by the field specified in KEY. This may be used to provide uniform sampling across allele frequencies, for instance.
vcfrocGenerates a pseudo-ROC curve using sensitivity and specificity estimated against a putative truth set. Thresholding is provided by successive QUAL cutoffs.
vcfstatsPrints statistics about variants in the input VCF file.
bFstbFst is a Bayesian approach to Fst. ImportantlybFst accounts for genotype uncertainty in the model using genotype likelihoods. For a more detailed description see: `A Bayesian approach to inferring population structure from dominant markers' by Holsinger et al. Molecular Ecology Vol 11, issue 7 2002. The likelihood function has been modified to use genotype likelihoods provided by variant callers. There are five free parameters estimated in the model: each subpopulation's allele frequency and Fis (fixation index, within each subpopulation), a free parameter for the total population's allele frequency, and Fst.
vcfcountallelesCount alleles
vcfparsealtsAlternate allele parsing method. This method uses pairwise alignment of REF and ALTs to determine component allelic primitives for each alternate allele.
genotypeSummaryGenerates a table of genotype counts. Summarizes genotype counts for bi-allelic SNVs and indel
vcfgenosummarizeAdds summary statistics to each record summarizing qualities reported in called genotypes. Uses: RO (reference observation count), QR (quality sum reference observations) AO (alternate observation count), QA (quality sum alternate observations)
vcfaltcountcount the number of alternate alleles in all records in the vcf file
meltEHH
iHSiHS calculates the integrated haplotype score which measures the relative decay of extended haplotype homozygosity (EHH) for the reference and alternative alleles at a site (see: voight et al. 2006, Spiech & Hernandez 2014).
segmentFstsegmentFst creates genomic segments (bed file) for regions with high wcFst
pFstpFst is a probabilistic approach for detecting differences in allele frequencies between two populations.
sequenceDiversityThesequenceDiversity program calculates two popular metrics of haplotype diversity: pi and extended haplotype homozygoisty (eHH). Pi is calculated using the Nei and Li 1979 formulation. eHH a convenient way to think about haplotype diversity. When eHH = 0 all haplotypes in the window are unique and when eHH = 1 all haplotypes in the window are identical.
vcfgenotypecompareadds statistics to the INFO field of the vcf file describing the amount of discrepancy between the genotypes (GT) in the vcf file and the genotypes reported in the . use this after vcfannotategenotypes to get correspondence statistics for two vcfs.
vcfrandomGenerate a random VCF file
segmentIhsCreates genomic segments (bed file) for regions with high wcFst
vcfgenotypesReport the genotypes for each sample, for each variant in the VCF. Convert the numerical represenation of genotypes provided by the GT field to a human-readable genotype format.

See alsovcflib.md.

scripts

The vcflib source repository contains a number of additional scripts.Click on the link to see the source code.

scriptdescription
vcfclearinfoclear INFO field
vcfqualfilterquality filter
vcfnulldotslashdotrewrite null genotypes to ./.
vcfprintaltdiscrepancy.rshow ALT discrepancies in a table
vcfremovenonATGCremove non-nucleotides in REF or ALT
plotSmoothed.Rsmooth plot of wcFst, pFst or abba-baba
vcf_strip_extra_headersstrip headers
plotHapLrt.Rplot results of pFst
vcfbiallelicremove anything that is not biallelic
vcfsortsort VCF using shell script
vcfnosnpsremove SNPs
vcfmultiwayscriptsmore multiway comparisons
vcfgtcompare.shannotates records in the first file with genotypes and sites from the second
plotPfst.Rplot pFst
vcfregionreduce_and_cutreduce, gzip, and tabix
plotBfst.Rplot results of pFst
vcfnobiallelicsnpsremove biallelic SNPs
vcfindelsshow INDELS
vcfmultiwaymultiway comparison
vcfregionreducereduce VCFs using a BED File, gzip them up and create tabix index
vcfprintaltdiscrepancy.shrunner
vcfclearidclear ID field
vcfcomplexremove all SNPs but keep SVs
vcffirstheadershow first header
plotXPEHH.Rplot XPEHH
vcfregionreduce_pipereduce, gzip and tabix in a pipe
vcfplotaltdiscrepancy.shplot ALT discrepancy runner
vcfplottstv.shrunner
vcfnoindelsremove INDELs
bgziptabixruns bgzip on the input and tabix indexes the result
plotHaplotypes.Rplot results
vcfplotsitediscrepancy.rplot site discrepancy
vcfindelproximityshow SNPs around an INDEL
bed2regionconvert VCF CHROM column in VCF file to region
vcfplotaltdiscrepancy.rplot ALT discrepancies
plot_roc.rplot ROC
vcfmultiallelicremove anything that is not multiallelic
vcfsnpsshow SNPs
vcfvarstatsuse fastahack to get stats
vcfregionreduce_uncompressedreduce, gzip and tabix
plotWCfst.Rplot wcFst
vcf2bed.pytransform VCF to BED file
vcfjoincallsoverlay files using QUAL and GT from a second VCF
vcf2sqlite.pypush VCF file into SQLite3 database using dbname

python

vcflib has rudimentary python bindings, but the are easy to build up on. Seepyvcflib.

Development

build from source

VCFLIB uses the cmake build system, after a recursive checkout of the sources make the files in the 'build' directory with:

git clone --recursive https://github.com/vcflib/vcflib.gitcd vcflibmkdir -p build&&cd buildcmake -DCMAKE_BUILD_TYPE=Debug -DZIG=OFF -DOPENMP=OFF ..cmake --build.cmake --install.

and to run the tests

ctest. --verbose

Executables are built into the./build directory in the repository.

Note, if you have an existing repo update submodules with

git submodule update --init --recursive --progresscd buildcmake --build. --target clean

The most interesting one, perhaps, is wfa2lib. Bundling is no longer the default, but it is a submodule that will install an optimized libwfa2.so. Set up compilation with

cmake -DCMAKE_BUILD_TYPE=Release -DWFA_GITMODULE=ON  ..cmake --build.ctest.cmake --install.

Build dependencies can be viewed in the github-CI scripts (see badges above), as well asguix.scm used by us to create the build environment (for instructions see the header of guix.scm). Essentially:

  • cmake
  • C++ compiler
  • htslib
  • tabixpp
  • WFA2 (wfa2lib)
  • pybind11 (for testing)

For include files add

  • libhts-dev
  • libtabixpp-dev
  • libtabixpp0

And for some of the VCF executables

  • zig 0.13 (disable with cmake -DZIG=OFF)
  • python
  • perl

To use Zig compiled tools, such as vcfcreatemulti, make sure a recentzig compiler is in thePATH.Thezig dependency is for the recently upgradedvcfcreatemulti that can be run after vcfwave to combine overlapping records. It is unfortunate that zig is not part of some large Linux distributions.

Note you need to add the full path before running cmake and make. From the build dir this should work:

cd buildexport PATH=$(pwd)/../zig-linux-x86_64-0.13.0:$PATHzig versioncmake  -DCMAKE_BUILD_TYPE=Debug -DWFA_GITMODULE=1 ..make VERBOSE=1

More onzig can be found in the souce codeREADME.

Using a recent C++ compiler

You may encounter something like libstdc++.so.6: version `GLIBCXX_3.4.32' not found (required by ...vcflib/build/pyvcflib.cpython-310-x86_64-linux-gnu.so). This is caused by the python pyvcflib not running against a python interpreter compiled with the same libstdc++ dependency. That is a bit nasty to solve. For now make sure to match C++ compilers for the python module.

Using a different htslib

Check out htslib in tabixpp (recursively) and

cmake -DHTSLIB_LOCAL:STRING=./htslib/ ..cmake --build .

link library

The standard build createsbuild/vcflib.a. Take a hint from thecmake file that builds all the vcflib tools.

distro builds

Distros, such as Debian, should build with something like

cmake -DCMAKE_BUILD_TYPE=RelWithDebInfo -DZIG=OFF ..

See the CMakeLists.txt header for more.

source code

Seevcfecho.cpp for basic usage.Variant.h andVariant.cppdescribe methods available in the API. vcflib is incorporated intoseveral projects, such asfreebayes, which may providea point of reference for prospective developers. Note vcflib containssubmodules (git repositories) comprising some dependencies. A fullGuix development environment we use is definedhere.

adding tests

vcflib uses different test systems. The most important one is thedoctest because itdoubles as documentation. For an example seevcf2tsv.md which can be run from thecommand line with

cdtestpython3 -m doctest -o NORMALIZE_WHITESPACE -o REPORT_UDIFF pytest/vcf2tsv.md

We also added support for python bindings and unit tests. Seerealign.py for an example.

Support

The developers are on the vcflibmatrix channel. Please do not use the github issue tracker for support issues!

Contributing

To contribute code to vcflib send a github pull request. We may askyou to add a working test case as described in 'adding tests'.

LICENSE

This software is distributed under the free softwareMIT LICENSE.

CREDIT

Citations are the bread and butter of Science. If you are using thissoftware in your research and want to support our future work, pleasecite the following publication:

Please cite:

A spectrum of free software tools for processing the VCF variant call format: vcflib, bio-vcf, cyvcf2, hts-nim and slivar.Garrison E, Kronenberg ZN, Dawson ET, Pedersen BS, Prins P (2022), PLoS Comput Biol 18(5): e1009123.https://doi.org/10.1371/journal.pcbi.1009123

Bibtex reference

@article{10.1371/journal.pcbi.1009123,doi ={10.1371/journal.pcbi.1009123},author ={Garrison, Erik AND Kronenberg, Zev N. AND Dawson, Eric T. AND Pedersen, Brent S. AND Prins, Pjotr},journal ={PLOS Computational Biology},publisher ={Public Library of Science},title ={A spectrum of free software tools for processing the VCF variant call format: vcflib, bio-vcf, cyvcf2, hts-nim and slivar},year ={2022},month ={05},volume ={18},url ={https://doi.org/10.1371/journal.pcbi.1009123},pages ={1-15}}

Below the prepublished version of our paper

@article {Garrison2021.05.21.445151,author ={Garrison, Erik and Kronenberg, Zev N. and Dawson, Eric T. and Pedersen, Brent S. and Prins, Pjotr},title ={Vcflib and tools for processing the VCF variant call format},elocation-id ={2021.05.21.445151},year ={2021},doi ={10.1101/2021.05.21.445151},publisher ={Cold Spring Harbor Laboratory},URL ={https://www.biorxiv.org/content/early/2021/05/23/2021.05.21.445151},eprint ={https://www.biorxiv.org/content/early/2021/05/23/2021.05.21.445151.full.pdf},journal ={bioRxiv}}

[8]ページ先頭

©2009-2025 Movatter.jp