vcftools: Usage and Options

The vcftools program is intended for analysis of diploid SNP data in VCF format. The program is run from the command line, and the interface is inspired by PLINK, and so should be largely familiar to users of that package. Commands take the following form:

vcftools --vcf file1.vcf --chr 20 --freq

The above command tells vcftools to read in the file file1.vcf, extract sites on chromosome 20, and calculate the allele frequency at each site. The resulting allele frequency estimates are stored in the output file, out.freq. As in the above example, output from vcftools is mainly sent to output files, as opposed to being shown on the screen.

If you want to output a new VCF file, you need the --recode option. For example, the command may look something like:

vcftools --vcf file1.vcf --chr 20 --out chr20 --recode

This command tells vcftools to extract chromosome 20 from file1.vcf and generate a new VCF file, which will be called chr20.recode.vcf.

A description of each of the available options follows. Note that some commands may only be available in the latest version of vcftools. To obtain the latest version, you should use SVN to checkout the latest code, as described on the home page.

Basic Options
Site Filter Options
Individual Filters
Genotype Filters
Output Statistics
Output in Other Formats
Miscellaneous
File Comparison
Options still in development

Basic Options

--vcf <filename>

vcftools expects files in VCF format v4.0, a specification of which can be found here

--gzvcf <filename>

--out <prefix>

This option defines the output filename prefix for all files generated by vcftools. For example, if <prefix> is set to output_filename, then all output files will be of the form output_filename.*** . If this option is omitted, all output files will have the prefix 'out.'.

Site Filter Options

--chr <chromosome>

Only process sites with a chromosome identifier matching <chromosome>
--from-bp <integer>
--to-bp <integer>

These options define the physical range of sites will be processed. Sites outside of this range will be excluded. These options can only be used in conjuction with --chr.
--snp <string>

Include SNP(s) with matching ID (e.g. a dbSNP rsID). This command can be used multiple times in order to include more than one SNP.
--snps <filename>

Include a list of SNPs given in a file. The file should contain a list of SNP IDs (e.g. dbSNP rsIDs), with one ID per line.
--exclude <filename>

Exclude a list of SNPs given in a file. The file should contain a list of SNP IDs, with one ID per line.
--positions <filename>
--exclude-positions <filename>

Include/exclude a set of sites on the basis of a list of positions in a file. Each line of the input file should contain a (tab-separated) chromosome and position. The file should have a header line.
--keep-only-indels
--remove-indels

Include or exclude sites that contain an indel. For this option 'indel' means any variant that alters the length of the REF allele.
--bed <filename>
--exclude-bed <filename>

Include or exclude a set of sites on the basis of a BED file. Only the first three columns (chrom, chromStart and chromEnd) are required. The BED file should have a header line.
--remove-filtered-all
--remove-filtered <string>
--keep-filtered <string>

These options are used to filter sites on the basis of their FILTER flag. The first option removes all sites with a FILTER flag. The second option can be used to exclude sites with a specific filter flag. The third option can be used to select sites on the basis of specific filter flags. The second and third options can be used multiple times to specify multiple FILTERs. The --keep-filtered option is applied before the --remove-filtered option.
--remove-INFO <string>
--keep-INFO <string>

These options are used to filter sites on the basis of INFO field flags. The first option can be used to exclude sites with a specific INFO flag. The second option can be used to select sites on the basis of specific INFO flags. These options can be used multiple times to specify multiple INFO flags. The --keep-INFO option is applied before the --remove-INFO option. Note that only INFO flags can currently be used as filters (i.e. there is currently no support for filtering by INFO field values).
--minQ <float>

Include only sites with Quality above this threshold.
--min-meanDP <float>
--max-meanDP <float>

Include sites with mean Depth within the thresholds defined by these options.
--maf <float>
--max-maf <float>

Include only sites with Minor Allele Frequency within the specified range.
--non-ref-af <float>
--max-non-ref-af <float>

Include only sites with all Non-Reference Allele Frequencies within the specified range.
--mac <int>
--max-mac <int>

Include only sites with Minor Allele Count within the specified range.
--non-ref-ac <float>
--max-non-ref-ac <float>

Include only sites with all Non-Reference Allele Counts within the specified range.
--hwe <float>

Assesses sites for Hardy-Weinberg Equilibrium using an exact test, as defined by Wigginton, Cutler and Abecasis (2005). Sites with a p-value below the threshold defined by this option are taken to be out of HWE, and therefore excluded.
--geno <float>

Exclude sites on the basis of the proportion of missing data (defined to be between 0 and 1, where 1 indicates no missing data allowed).
--max-missing-count <int>

Exclude sites with more than this number of missing chromosomes.
--min-alleles <int>
--max-alleles <int>

Include only sites with a number of alleles within the specified range. For example, to include only bi-allelic sites, one could use:

vcftools --vcf file1.vcf --min-alleles 2 --max-alleles 2
--thin <int>

Thin sites so that no two sites are within the specified distance.
--mask <filename>
--invert-mask <filename>
--mask-min <int>

Include sites on the basis of a FASTA-like file. The provided file contains a sequence of integer digits (between 0 and 9) for each position on a chromosome that specify if a site at that position should be filtered or not. An example mask file would look like:

>1
0000011111222...
In this example, sites in the VCF file located within the first 5 bases of the start of chromosome 1 would be kept, whereas sites at position 6 onwards would be filtered out. The threshold integer that determines if sites are filtered or not is set using the --mask-min option, which defaults to 0. The chromosomes contained in the mask file must be sorted in the same order as the VCF file. The --mask option is used to specify the mask file to be used, whereas the --invert-mask option can be used to specify a mask file that will be inverted before being applied.

Individual Filters

--indv <string>

Specify an individual to be kept in the analysis. This option can be used multiple times to specify multiple individuals.
--keep <filename>

Provide a file containing a list of individuals to include in subsequent analysis. Each individual ID (as defined in the VCF headerline) should be included on a separate line.
--remove-indv <string>

Specify an individual to be removed from the analysis. This option can be used multiple times to specify multiple individuals. If the --indv option is also specified, then the --indv option is executed before the --remove-indv option.
--remove <filename>

Provide a file containing a list of individuals to exclude in subsequent analysis. Each individual ID (as defined in the VCF headerline) should be included on a separate line. If both the --keep and the --remove options are used, then the --keep option is execute before the --remove option.
--min-indv-meanDP <float>
--max-indv-mean-DP <float>

Calculate the mean coverage on a per-individual basis. Only individuals with coverage within the range specified by these options are included in subsequent analyses.
--mind <float>

Specify the minimum call rate threshold for each individual.
--phased

First excludes all individuals having all genotypes unphased, and subsequently excludes all sites with unphased genotypes. The remaining data therefore consists of phased data only.
--max-indv <int>

Randomly thins individuals so that only the specified number are retained.

Genotype Filters

--remove-filtered-geno-all
--remove-filtered-geno <string>

The first option removes all genotypes with a FILTER flag. The second option can be used to exclude genotypes with a specific filter flag.
--minGQ <float>

Exclude all genotypes with a quality below the threshold specified by this option (GQ).
--minDP <float>
--maxDP <float>

Exclude all genotypes with a sequencing depth (DP) outside the range defined by these options.

Output Statistics

--freq
--counts
--freq2
--counts2

Output per-site frequency information. The --freq outputs the allele frequency in a file with the suffix '.frq'. The --counts option outputs a similar file with the suffix '.frq.count', that contains the raw allele counts at each site.
The --freq2 and --count2 options are used to suppress allele information in the output file. In this case, the order of the freqs/counts depends on the numbering in the VCF file.
--depth

Generates a file containing the mean depth per individual. This file has the suffix '.idepth'.
--site-depth
--site-mean-depth

Generates a file containing the depth per site. The --site-depth option outputs the depth for each site summed across individuals. This file has the suffix '.ldepth'. Likewise, the --site-mean-depth outputs the mean depth for each site, and the output file has the suffix '.ldepth.mean'.
--geno-depth

Generates a (possibly very large) file containing the depth for each genotype in the VCF file. Missing entries are given the value -1. The file has the suffix '.gdepth'.
--site-quality

Generates a file containing the per-site SNP quality, as found in the QUAL column of the VCF file. This file has the suffix '.lqual'.
--het

Calculates a measure of heterozygosity on a per-individual basis. Specfically, the inbreeding coefficient, F, is estimated for each individual using a method of moments. The resulting file has the suffix '.het'.
--hardy

Reports a p-value for each site from a Hardy-Weinberg Equilibrium test (as defined by Wigginton, Cutler and Abecasis (2005)). The resulting file (with suffix '.hwe') also contains the Observed numbers of Homozygotes and Heterozygotes and the corresponding Expected numbers under HWE.
--missing

Generates two files reporting the missingness on a per-individual and per-site basis. The two files have suffixes '.imiss' and '.lmiss' respectively.
--hap-r2
--geno-r2
--ld-window <int>
--ld-window-bp <int>
--min-r2 <float>

These options are used to report Linkage Disequilibrium (LD) statistics as summarised by the r², D, and D' statistics. The --hap-r2 option informs vcftools to output a file reporting the r², D, and D' statistics using phased haplotypes. These are the traditional measures of LD often reported in the population genetics literature. If phased haplotypes are unavailable then the --geno-r2 option may be used, which calculates the squared correlation coefficient between genotypes encoded as 0, 1 and 2 to represent the number of non-reference alleles in each individual. This is the same as the LD measure reported by PLINK. The D and D' statistics are only available for phased genotypes. The haplotype version outputs a file with the suffix '.hap.ld', whereas the genotype version outputs a file with the suffix '.geno.ld'. The haplotype version implies the option --phased.

The --ld-window option defines the maximum number of SNPs between the SNPs being tested for LD. Likewise, the --ld-window-bp option can be used to define the maximum physical separation (in base-pairs) of SNPs included in the LD calculation. Finally, the --min-r2 sets a minimum value for r² below which the LD statistic is not reported.
--SNPdensity <int>

Calculates the number and density of SNPs in bins of size defined by this option. The resulting output file has the suffix '.snpden'.
--TsTv <int>

Calculates the Transition / Transversion ratio in bins of size defined by this option. Only uses bi-allelic SNPs. The resulting output file has the suffix '.TsTv'. A summary is also supplied in a file with the suffix '.TsTv.summary'.
--TsTv-by-count

Calculates the Transition / Transversion ratio as a function of alternative allele count. Only uses bi-allelic SNPs. The resulting output file has the suffix '.TsTv.count'.
--TsTv-by-qual

Calculates the Transition / Transversion ratio as a function of SNP quality threshold. Only uses bi-allelic SNPs. The resulting output file has the suffix '.TsTv.qual'.
--FILTER-summary

Generates a summary of the number of SNPs and Ts/Tv ratio for each FILTER category. The output file has the suffix '.FILTER.summary.
--filtered-sites

Creates two files listing sites that have been kept or removed after filtering. The first file, with suffix '.kept.sites', lists sites kept by vcftools after filters have been applied. The second file, with the suffix '.removed.sites', list sites removed by the applied filters.
--singletons

This option will generate a file detailing the location of singletons, and the individual they occur in. The file reports both true singletons, and private doubletons (i.e. SNPs where the minor allele only occurs in a single individual and that individual is homozygotic for that allele). The output file has the suffix '.singletons'.
--site-pi
--window-pi <int>

These options are used to estimate levels of nucleotide diversity. The first option does this on a per-site basis, and the output file has the suffix '.sites.pi'. The second option calculates the nucleotide diversity in windows, with the window size defined in the option argument.

Output in Other Formats

--012

This option outputs the genotypes as a large matrix. Three files are produced. The first, with suffix '.012', contains the genotypes of each individual on a separate line. Genotypes are represented as 0, 1 and 2, where the number represent that number of non-reference alleles. Missing genotypes are represented by -1. The second file, with suffix '.012.indv' details the individuals included in the main file. The third file, with suffix '.012.pos' details the site locations included in the main file.
--IMPUTE

This option outputs phased haplotypes in IMPUTE reference-panel format. As IMPUTE requires phased data, using this option also implies --phased. Unphased individuals and genotypes are therefore excluded. Only bi-allelic sites are included in the output. Using this option generates three files. The IMPUTE haplotype file has the suffix '.impute.hap', and the IMPUTE legend file has the suffix '.impute.hap.legend'. The third file, with suffix '.impute.hap.indv', details the individuals included in the haplotype file, although this file is not needed by IMPUTE.
--ldhat
--ldhat-geno

These options output data in LDhat format. Use of these options also require the --chr option to by used. The --ldhat option outputs phased data only, and therefore also implies --phased, leading to unphased individuals and genotypes being excluded. Alternatively, the --ldhat-geno option treats all of the data as unphased, and therefore outputs LDhat files in genotype/unphased format. In either case, two files are generated with the suffixes '.ldhat.sites' and '.ldhat.locs', which correspond to the LDhat 'sites' and 'locs' input files respectively.
--BEAGLE-GL

This option outputs genotype likelihood information for input into the BEAGLE program. This option requires the VCF file to contain the FORMAT GL tag, which can generally be output by SNP callers such as the GATK. Use of this option requires a chromosome to be specified via the --chr option. The resulting output file (with the suffix '.BEAGLE.GL') contains genotype likelihoods for biallelic sites, and is suitable for input into BEAGLE via the 'like=' argument.
--plink

This option outputs the genotype data in PLINK PED format. Two files are generated, with suffixes '.ped' and '.map'. Note that only bi-allelic loci will be output. Further details of these files can be found in the PLINK documentation.

Note: This option can be very slow on large datasets. Using the --chr option to divide up the dataset is advised, or alternatively use the --plink-tped option below.
--plink-tped

The --plink option above can be extremely slow on large datasets. An alternative that might be considerably quicker is to output in the PLINK transposed format. This can be achieved using the --plink-tped option, which produces two files with suffixes '.tped' and '.tfam'.
--recode

The --recode option is used to generate a VCF file from the input VCF file having applied the options specified by the user. The output file has the suffix '.recode.vcf'.

By default, the INFO fields are removed from the output file, as the INFO values may be invalidated by the recoding (e.g. the total depth may need to be recalculated if individuals are removed). This default functionality can be overridden by using the --recode-INFO <string> option, where <string> defines the INFO key to keep in the output file. The --recode-INFO flag can be used multiple times. Alternatively, the option --recode-INFO-all can be used to retain all INFO fields.
--recode-to-stream

The --recode-to-stream option works in the same manner as the --recode option, but the resulting VCF file printed directly to the screen, rather than to file. Logging information normally displayed on the screen is supressed (but still written to the log file). The advantage of this option is that the resulting VCF file can be piped to other programs for processing. For example, to gzip the VCF file, one could use:

vcftools --vcf file1.vcf --recode-to-stream | gzip -c - > output_file.vcf.gz

Miscellaneous

--extract-FORMAT-info <string>

Extract information from the genotype fields in the VCF file relating to a specfied FORMAT identifier. For example, using the option '--extract-FORMAT-info GT' would extract the all of the GT (i.e. Genotype) entries. The resulting output file has the suffix '.<FORMAT_ID>.FORMAT'.
--get-INFO <string>

This option is used to extract information from the INFO field in the VCF file. The <string> argument specifies the INFO tag to be extracted, and the option can be used multiple times in order to extract multiple INFO entries. The resulting file, with suffix '.INFO', contains the required INFO information in a tab-separated table. For example, to extract the NS and DB flags, one would use the command:

vcftools --vcf file1.vcf --get-INFO NS --get-INFO DB
--force-index-write

When vcftools first reads a VCF file, it creates an index of the file for faster access next time. In some cases, the index file can get out of sync with the VCF file. This option can be used to force vcftools to regenerate the index file.

VCF File Comparison Options

The file comparison options are currently in a state of flux and likely buggy. If you find a bug, please report it. Note that genotype-level filters are not supported in these options.

--diff <filename>
--gzdiff <filename>

Select a VCF file for comparison with the file specified by the --vcf option. Outputs two files describing the sites and individuals common / unique to each file. These files have the suffixes '.diff.sites_in_files' and '.diff.indv_in_files' respectively. The --gzdiff version can be used to read compressed VCF files.
--diff-site-discordance

Used in conjuction with the --diff option to calculate discordance on a site by site basis. The resulting output file has the suffix '.diff.sites'.
--diff-indv-discordance

Used in conjuction with the --diff option to calculate discordance on a per-individual basis. The resulting output file has the suffix '.diff.indv'.
--diff-discordance-matrix

Used in conjuction with the --diff option to calculate a discordance matrix. This option only works with bi-allelic loci with matching alleles that are present in both files. The resulting output file has the suffix '.diff.discordance.matrix'.
--diff-switch-error

Used in conjuction with the --diff option to calculate phasing errors (specifically 'switch errors'). This option generates two output files describing switch errors found between sites, and the average switch error per individual. These two files have the suffixes '.diff.switch' and '.diff.indv.switch' respectively.

Options still in development

The following options are yet to be finalised, are likely to contain bugs, and are likely to change in the future.

--hapmap-fst-file <filename>
--hapmap-fst-gzfile <filename>

Calculate F_ST for a pair of VCF files, with the second file being specified by this option. F_ST is currently calculated using the formula described in the supplementary material of the Phase I HapMap paper. Currently, only pairwise F_ST calculations are supported, although this will likely change in the future. The --gzfst option can be used to read compressed VCF files. This option is likely to be depreciated in future releases, and it is recommended that you use the --hapmap-fst-pop or --weir-fst-pop options instead.
--hapmap-fst-pop <filename>
--weir-fst-pop <filename>

These options are used to calculate two estimates of F_ST. The --hapmap-fst-pop option is used to calculate the estimator described in the Phase I HapMap paper. The second (preferred) estimator is from Weir and Cockerham's 1984 paper. These options are used to provide a file which lists the individuals in a given population, and is used multiple times to specific multiple populations. These options can be used to calculate F_ST for two or more populations by re-using the option to specify each population file.
--LROH

Identify Long Runs of Homozygosity.
--relatedness

Output Individual Relatedness Statistics.
--TajimaD <int>

Output Tajima's D statistic in bins of size <int>.

vcftools: Usage and Options

Navigation