/GangSTR

A tool for profiling long STRs from short reads

Primary LanguageC++OtherNOASSERTION

GangSTR

GangSTR

GangSTR is a tool for genome-wide profiling tandem repeats from short reads. A key advantage of GangSTR over existing genome-wide TR tools (e.g. lobSTR or hipSTR) is that it can handle repeats that are longer than the read length.

GangSTR takes aligned reads (BAM) and a set of repeats in the reference genome as input and outputs a VCF file containing genotypes for each locus.

Manuscript: https://doi.org/10.1093/nar/gkz501

For questions on installation or usage, please open an issue, submit a pull request, or contact Nima Mousavi (mousavi@ucsd.edu).

For advanced topics such as those below, see the GangSTR wiki.

A Docker with GangSTR plus the dumpSTR filtering tool installed is available at gymreklab/str-toolkit from Docker hub.

Download | Install | Basic Usage | File formats | Reference files

Download

The latest GangSTR release is available on the releases page.

For a list of TR references available, see references below.

Prerequisites

  • A recent version of C/C++ compiler supporting C++11 standard
  • CMake version 3.16 or above
  • The following development files in the build system: libz-dev, libbz2-dev, and liblzma-dev (required by htslib)

Basic Install

If you are installing from the tarball (which for most purposes you should be), the following instructions will install all dependencies as well as GangSTR itself. Both UNIX and Mac OSX are supported.

If you are running as root:

tar -xzvf GangSTR-X.X.tar.gz
cd GangSTR-X.X
mkdir build
cd build
cmake ..
make
sudo cmake --install .

If you are installing locally (e.g. on a cluster where you don't have root access):

tar -xzvf GangSTR-X.X.tar.gz
cd GangSTR-X.X
mkdir build
cd build
cmake ..
make
cmake --install . --prefix PREFIX

where PREFIX is a place you have write permissions. In most cases this will be your home directory, e.g. $HOME. If you install locally, make sure $PREFIX/bin is on your PATH.

Typing GangSTR --help should show a help message if GangSTR was successfully installed.

Compiling from git source

To compile from git source:

# Clone the repo
git clone https://github.com/gymreklab/GangSTR
cd GangSTR/
mkdir build
cmake ..
make
cmake --install . --prefix PREFIX

Install using conda

You can install GangSTR v2.5.0 using conda (or mamba) package manager.

conda install -c bioconda -c conda-forge gangstr

Special thanks to the users in this thread for help in setting this up.

Basic usage

To run GangSTR using default parameters use the following command:

GangSTR --bam file.bam 
        --ref ref.fa 
        --regions regions.bed 
        --out outprefix 

Required parameters:

  • --bam <file.bam,[file2.bam]> Comma separated list of input BAM files
  • --ref Refererence genome (.fa)
  • --regions Target TR loci (regions) (.bed)
  • --out Output prefix

Additional general options:

  • --targeted Run GangSTR in targeted mode. This mode should be used when targeting disease loci. (as opposed to genome-wide run)
  • --chrom <string> Only genotype regions on this chromosome.
  • --bam-samps <string> Comma separated list of sample IDs for --bam
  • --samp-sex <string> Comma separated list of sample sex for each sample ID (--bam-samps must be provided, see readme for more details)
  • --str-info <string> Tab file with additional per-STR info (e.g., expansion cutoff. see below for format)
  • --period <string> Only genotype loci with periods (motif lengths) in this comma-separated list.
  • --skip-qscore Skip calculation of Q-score (see Q field in VCF output).

Options for different sequencing settings

  • --readlength <int> Preset read length (default: extract from alignments if not provided)
  • --coverage <float> Preset average coverage, should be set for exome/targeted data. Comma separated list to specify for each BAM. (default: calculate if not provided)
  • --model-gc-coverage Model coverage as a function of GC content. Requires genome-wide data. Experimental feature.
  • --insertmean <float> Fragment length mean. (default: calculate if not provided)
  • --insertsdev <float> Fragment length standard deviation. (default: calculate if not provided)
  • --nonuniform Indicates non-uniform coverage in alignment file (i.e., used for exome sequencing). Using this flag removes the likelihood term corresponding to FRR count.
  • --min-sample-reads <int> Minimum number of reads per sample.

Advanced parameters for likelihood model:

  • --frrweight <float> Reset weight for FRR class in likelihood model. (default 1.0)
  • --spanweight <float> Reset weight for Spanning class in likelihood model. (default 1.0)
  • --enclweight <float> Reset weight for Enclosing class in likelihood model. (default 1.0)
  • --flankweight <float> Reset weight for Flanking class in likelihood model. (default 1.0)
  • --ploidy [1,2] Haploid (1) or diploid (2) genotyping. (default 2)
  • --skipofftarget Skip off target regions included in the regions file.
  • --readprobmode Only use read probabilities in likelihood model. (ignore class probability)
  • --numbstrap <int> Number of bootstrap samples for calculating confidence intervals. (default 100)
  • --grid-theshold <int> Use optimization rather than grid search to find MLE if search space (grid) contains more alleles than this threshold. Default: 10000
  • --rescue-count <int> Number of regions that GangSTR attempts to rescue mates from (excluding off-target regions). Default: 0
  • --max-proc-read <int> Maximum number of processed reads per sample before a region is skipped.

Parameters for local realignment:

  • --minscore <int> Minimun alignment score for accepting reads (default 75).
  • --minmatch <int> Minimum matching basepairs required at the edge of the repeat region to accept flanking and enclosing reads (default 5).

Stutter model parameters:

  • --stutterup <float> Stutter insertion probability (default 0.05)
  • --stutterdown <float> Stutter deletion probability (default: 0.05)
  • --stutterprob <float> Stutter step size parameter (default: 0.90)

Parameters for more detailed info about each locus:

  • --output-readinfo Output a file containing extracted read information.
  • --output-bootstraps Output a file containing bootstrap samples.
  • --include-ggl Output GGL (special GL field) in VCF.

Additional optional parameters:

  • -h,--help display help screen
  • --quiet Don't print out anything
  • --seed Random number generator initial seed
  • -v,--verbose Print progress information (major steps)
  • --very Print detailed progress information
  • --version Print out the version of this software

File formats

GangSTR takes as input a BAM file of short read alignments, a reference set of TRs, and a reference genome, and outputs genotypes in a VCF file. Each of these formats is described below.

BAM (--bam)

GangSTR requires a BAM file produced by an indel-sensitive aligner. The BAM file must be sorted and indexed e.g. by using samtools sort and samtools index. GangSTR currently only processes a single sample at a time.

FASTA Reference genome (--ref)

You must input a reference genome in FASTA format. This must be the same reference build used to align the sequences in the BAM file.

TR regions (--regions)

GangSTR requires a reference set of regions to genotype. This is a BED-like file with the following columns:

  1. The name of the chromosome on which the STR is located
  2. The start position of the STR on its chromosome
  3. The end position of the STR on its chromosome
  4. The motif length
  5. The repeat motif

An optional 6th column may contain a comma-separated list of off-target regions for each TR. These are regions where misaligned reads for a given TR may be incorrectly mapped.

Below is an example file which contains 5 TR loci. Standard references for hg19 and GRCh38 can be obtained below. NOTE: The table header is for descriptive purposes. The BED file should not have a header

CHROM START END MOTIF_LEN MOTIF OFFTARGET (optional)
chr1 10689 10700 5 CGCGC
chr1 28589 28603 1 T
chr4 11173 11194 11 CGCCGGCGCGG
chr4 150889 150909 2 TG
chr19 45770205 45770264 3 CAG chr2:163338502-163338506,chr3:197333949-197333955,chr6:16327632-16327646,chr6:170561926-170561931,chr7:122288209-122288215,chr8:133055822-133055827,chr11:28310883-28310888,chr17:4887671-4887677,chr18:55586148-55586165,chr19:13207866-13207871

--str-info

A tab delimited with the following header and format can be used to specify additional per locus information. GangSTR currently supports expansion threshold through str-info. The threshold is specified in number of repeat copies, and it is used to calculate expansion probability. (See QEXP field in VCF format). Note: The loci represented in this file are unique and duplicates should be removed.

chrom pos end thresh
chr1 26454 26465 50
chr1 31556 31570 20
chr1 35489 35504 25

VCF (output)

For more information on VCF file format, see the VCF spec. In addition to standard VCF fields, GangSTR adds custom fields described below.

INFO fields

INFO fields contain aggregated statistics about each TR. The following custom fields are added:

FIELD DESCRIPTION
END End position of the TR
PERIOD Length of the repeat unit
GRID Range of the optimization grid. Gives min and max repeat copy number considered
EXPTHRESH The threshold copy number used to test for repeat expansions
STUTTERUP The model probability to observe a stutter error increasing the repeat number
STUTTERDOWN The model probability to observe a stutter error decreasing the repeat number
STUTTERP The geometric parameter for modeling the stutter step size distribution
RU Repeat motif
REF Reference copy number (number of repeat units

FORMAT fields

FORMAT fields contain information specific to each genotype call. The following custom fields are added:

FIELD DESCRIPTION
GT Genotype
DP Read Depth (number of informative reads)
Q Quality Score
REPCN Genotype given in number of copies of the repeat motif
REPCI 95% Confidence interval for each allele based on bootstrapping
RC Number of reads in each class (enclosing, spanning, FRR, flanking)
ENCLREADS Summary of reads in enclosing class in | separated key-value pairs. Keys are number of copies and values show number of reads with that many copies.
FLNKREADS Summary of reads in flanking class in | separated key-value pairs. Keys are number of copies and values show number of reads with that many copies.
ML Maximum likelihood
INS Insert size mean and stddev at the locus
STDERR Bootstrap standard error of each allele
QEXP Prob. of no expansion, 1 expanded allele, both expanded alleles
GGL Genotpye Likelihood of all pairs of alleles in the search space. Formatted similar to standard GL fields but with allele space defined by the INFO/GRID field

Q: Quality score estimated alleles (REPCN), between 0 and 1. This quality score is a measure of GangSTR's confidence in short allele calls (shorter than read length). It gives the likelihood of the maximum likelihood genotype divided by the sum of likelihoods of all possible genotypes. This can be interpreted as a posterior probability with a uniform prior over all possible genotypes. Calculation of Q-score can be slow if the estimation search space (grid) is large. To skip this step, use --skip-qscore option.

STDERR: Standard error of estimated alleles using bootstrap method.

QEXP: Given estimated alleles, the likelihood plane, and an expansion threshold, this field shows three numbers: the probability of both alleles being smaller than the threshold, one allele larger and one smaller than threshold, and both alleles larger than threshold. The expansion threshold should be provided using --str-info field.

Read info file (output)

By using --output-readinfo a file with .readinfo.tab extention containing information from the reads extracted for each locus is generated. The columns are ordered as follows:

Column number Description
1 Chromosome
2 Repeat start position
3 Repeat end position
4 Read ID (originated from BAM file)
5 Read class {**}
6 Read class data field {**}
7 Found mate (boolean flag)

{**} Read class codes and their corresponding data field

Each read in the .readinfo.tab file belongs to one of 5 classes. The following table shows what each read class code means and how to interpret the read class data field column. For more information on read classes please refer to manuscript https://doi.org/10.1093/nar/gkz501.

Read Class Code Description Data field
SPAN Spanning read pair Fragment length (insert size) of the spanning read pair
SPFLNK A flanking read that creates a spanning read pair with its mate Number of repeat copies on the flanking read
BOUND A flanking read Number of repeat copies on the flanking read
ENCLOSE Enclosing read Number of repeat copies enclosed in the read
FRR Fully repetitive read Distance of mate from the repeat region (set to -(read_length) if mate is also FRR)

GangSTR reference files

The following lists available references created using Tandem Repeats Finder. We update the reference periodically with additional loci or annotation changes. Note references must be unzipped before using with GangSTR. The file listed in bold is the current recommended version.

Reference build Version Link Comment
hg19 ver13.1 hg19_ver13_1.bed.gz Added two disease loci to the reference (C9ORF72 and FMR1)
hg19 ver13 hg19_ver13.bed.gz More strict removal of locus bundles
hs37 ver13 hs37_ver13.bed.gz More strict removal of locus bundles
hg38 ver13 hg38_ver13.bed.gz More strict removal of locus bundles
hg19 ver12 hg19_ver12.bed.gz Motifs of up to 20 bps, includes ChrX and ChrY, Trimmed messy loci
hs37 ver12 hs37_ver12.bed.gz Motifs of up to 20 bps, includes ChrX and ChrY, Trimmed messy loci
hg38 ver12 hg38_ver12.bed.gz Motifs of up to 20 bps, Trimmed messy loci
hg19 ver8 hg19_ver8.bed.gz Motifs of up to 15 bps
hg19 ver10 hg19_ver10.sorted.bed.gz Motifs of up to 20 bps, includes ChrX and ChrY
hs37 ver8 hs37_ver8.bed.gz Motifs of up to 15 bps
hg38 ver5 hg38_ver5.bed.gz Motifs of up to 15 bps
hg38 ver6 hg38_ver6.sorted.bed.gz Motifs of up to 20 bps, includes ChrX and ChrY
hg38 ver16 hg38_ver16.bed.gz
hg38 ver17 hg38_ver17.bed.gz

The references below contain pre-defined off-target loci for target pathogenic loci (hg38 coordinates):

Locus hg38 Link hg19 Link
SCA1 SCA1_hg38.bed SCA1_hg19.bed
SCA2 SCA2_hg38.bed SCA2_hg19.bed
SCA3 SCA3_hg38.bed SCA3_hg19.bed
SCA6 SCA6_hg38.bed SCA6_hg19.bed
SCA7 SCA7_hg38.bed SCA7_hg19.bed
SCA8 SCA8_hg38.bed SCA8_hg19.bed
SCA12 SCA12_hg38.bed SCA12_hg19.bed
SCA17 SCA17_hg38.bed SCA17_hg19.bed
HTT HTT_hg38.bed HTT_hg19.bed
DM1 DM1_hg38.bed DM1_hg19.bed
FMR1 FMR1_hg38.bed FMR1_hg19.bed
C9ORF72 C9ORF72_hg38.bed C9ORF72_hg19.bed

Non-human reference builds:

Reference build Version Link Comment
mm10 ver2 mm10_ver2.bed.gz
mm9 ver2 mm9_ver2.bed.gz

GangSTR callsets

GangSTR callsets on publicly available datasets.

Dataset Reference build and version Link
NA12878, NA12891, NA12892 hg19 v13.1 NA12878_trio_hg19_v13_1_filtered_level1.vcf.gz

Calling on sex chromosomes.

You can call TRs on chrX and chrY using a combination of --bam-samps and --samp-sex. --samp-sex is a list of sex assignments ('F' or 'M') for the list of samples in --bam-samps, in the same order. For example if sample1 and sample2 are Male and Female respectively, --bam-samps sample1,sample2 --samp-sex M,F as input option.

Currently, GangSTR is not capable of extracting sample sex automatically.