SVTrek

SVTrek is a bioinformatics tool for auditing structural variation (SV) discoveries and can also perform SV discovery independently using long-read analysis.

Dependencies

• A C compiler (gcc or clang) and make
• autoconf, automake, and libtool (only needed when htslib is built locally, for autoreconf)
• zlib development headers
• pkg-config (used to detect a system htslib)

Installation

# clone the repository
git clone --recursive https://github.com/akmami/SVTrek.git
cd SVTrek

# build dependencies
make install

# compile the program
make

If you already cloned without --recursive, initialise the submodule first:

git submodule update --init --recursive

Choosing the htslib source

make install (and make) accept a HTSLIB variable:

• make install HTSLIB=auto — use system htslib if pkg-config finds it, otherwise build locally (this is the default).
• make install HTSLIB=system — require the system htslib reported by pkg-config.
• make install HTSLIB=local — always clone + build htslib under deps/htslib, ignoring any system copy.

Use the same HTSLIB=... value for the subsequent make so the compile and link flags match.

Other make targets

• make clean — remove the svtrek binary and object files.
• make distclean — also remove the locally built deps/htslib (a system htslib and the abPOA submodule are left untouched).

Usage

./svtrek [MODE] [OPTIONS]

Program Modes

disc

SV discovery mode on a graph-alignment result. It takes a GFA reference graph, a GAF alignment, and the FASTQ reads.

Usage

./svtrek disc [-r|--gfa GFA] [-a|--gaf GAF] [-q|--fq FASTQ] [OPTIONS]

Required Parameters

• -r, --gfa <GFA>
  • Reference graph in GFA format.
• -a, --gaf <GAF>
  • Graph alignment in GAF format.
• -q, --fq <FASTQ>
  • Long reads in FASTQ format.

Options

• -o, --output <filename>
  • Output filename.
  • Default: svtrek.out
• -t <num>
  • Number of threads to use for processing.
  • Default: 4
• --verbose
  • Enables verbose output.
  • Default: false
• --consensus-interval-range <num>
  • Interval that limits the refinement range.
  • Default: 500
• --consensus-interval <num>
  • Interval within which locations are treated as the same position.
  • Default: 5
• --consensus-min-count <num>
  • Minimum number of elements required for consensus determination.
  • Default: 3

audt

Variation auditing mode: parses a BAM file and validates the variations reported in a VCF file.

Usage

./svtrek audt [-b|--bam BAM] [-v|--vcf VCF] [OPTIONS]

Required Parameters

• -b, --bam <BAM>
  • BAM file to be processed.
• -v, --vcf <VCF>
  • VCF file to be audited.

Options

• -o, --output <filename>
  • Output filename.
  • Default: svtrek.out
• -t <num>
  • Number of threads to use for processing.
  • Default: 4
• --verbose
  • Enables verbose output.
  • Default: false
• --wider-interval <num>
  • Offset interval for the start of the reads (DEL-START).
  • Default: 20000
• --median-interval <num>
  • Offset interval for point positions (INS).
  • Default: 10000
• --narrow-interval <num>
  • Offset interval for the end of the reads (DEL-END).
  • Default: 2000
• --consensus-interval-range <num>
  • Interval that limits the refinement range.
  • Default: 500
• --consensus-interval <num>
  • Interval within which reads are treated as being at the same position.
  • Default: 5
• --consensus-min-count <num>
  • Minimum number of elements required for consensus determination.
  • Default: 3

Simulation & benchmarking

Helper scripts for generating a simulated benchmark live under misc/:

• misc/simulate.sh — simulates SVs/SNPs on a reference (SURVIVOR), generates long reads (Badread), and maps them (minimap2), producing a mutated genome, a gold-standard VCF, and a sorted BAM. Species profiles are in misc/params/.
• misc/fix_survivor_vcf.awk — normalises SURVIVOR's VCF output into a valid, tab-delimited VCF that bcftools/truvari accept.