MALVA: genotyping by Mapping-free ALternate-allele detection of known VAriants

Alignment-free genotyping of a set of known variants (in VCF format) directly from a sample of reads.

Install

MALVA is available on bioconda.

$ conda create -n malvatest -c bioconda malva

will create an environment named malvatest that includes MALVA and its dependencies.

Install from source code

Dependencies

To manually compile MALVA you'll need a C++17-compliant compiler, CMake, and the following libraries installed in your system:

• sdsl-lite v2.1.1 under GPLv3 License
• KMC >= v2.3 under GPLv3 License
• htslib >= v1.10.2 under MIT/Expat License
• zstd >= 1.5.1 under BSD License and GPLv2
• zstdstream (included in this repository) under MIT License
• xxHash (included in this repository) under BSD 2-Clause License
• zlib under zlib license

Use your favorite system-wide package manager to install them before compiling MALVA.

Alternatively, you can also use conda (and bioconda) to install dependencies. For example (please adapt to you system setup):

conda create -n malvadeps -c conda-forge -c bioconda htslib kmc sdsl-lite cmake zstd-static cxx-compiler

Notice that these dependencies are needed only if you want to compile MALVA from sources, since otherwise it is already available on Bioconda in binary form (see above).

Download and installation

To download and compile the code run the following commands.

git clone https://github.com/AlgoLab/malva.git
cd malva
mkdir build
cd build
cmake -DCMAKE_BUILD_TYPE=Release ..
make

If the compilation is successful, the malva-geno binary will be copied to the ${PROJECT_ROOT}/bin directory.

Usage

Usage: malva-geno <subcommand> [-k KMER-SIZE] [-r REF-KMER-SIZE] [-c MAX-COV] <reference> <variants> <kmc_output_prefix>

Arguments:
	<subcommand>                      either index to create the reference index or call to call call the genotypes.

    -h, --help                        display this help and exit
    -k, --kmer-size                   size of the kmers to index (default:35)
    -r, --ref-kmer-size               size of the reference kmers to index (default:43)
    -e, --error-rate                  expected sample error rate (default:0.001)
    -s, --samples                     file containing the list of (VCF) samples to consider (default:-, i.e. all samples)
    -f, --freq-key                    a priori frequency key in the INFO column of the input VCF (default:AF)
    -c, --max-coverage                maximum coverage for variant alleles (default:200)
    -b, --bf-size                     bloom filter size in GB (default:4)
    -p, --strip-chr                   strip "chr" from sequence names (default:false)
    -u, --uniform                     use uniform a priori probabilities (default:false)
    -v, --verbose                     output COVS and GTS in INFO column (default: false)
    -1, --haploid                     run MALVA in haploid mode (default: false)

Positional arguments:
    <reference>                       reference file in FASTA format (may be gzipped)
    <variants>                        variants file in VCF format (may be gzipped)
    <kmc_output_prefix>               prefix of KMC output

The file needed by malva whose prefix is <kmc_output_prefix> can be computed with KMC as follows:

kmc -k<REF-KMER-SIZE> <sample> <kmc_output_prefix> <kmc_tmp_dir>

Anyway, we provide a bash script that you can use to run the full pipeline KMC+malva-geno:

Usage: MALVA [-k KMER-SIZE] [-r REF-KMER-SIZE] [-c MAX-COV] <reference> <variants> <sample>

Arguments:
     -h              print this help and exit
     -k              size of the kmers to index (default:35)
     -r              size of the reference kmers to index (default:43)
     -e              expected sample error rate (default:0.001)
     -s              file containing the list of (VCF) samples to consider (default:-, i.e. all samples)
     -f              a priori frequency key in the INFO column of the input VCF (default:AF)
     -c              maximum coverage for variant alleles (default:200)
     -b              bloom filter size in GB (default:4)
     -m              max amount of RAM in GB - KMC parameter (default:4)
     -p              strip "chr" from sequence names (dafault:false)
     -u              use uniform a priori probabilities (default:false)
     -v              output COVS and GTS in INFO column (default: false)
     -1              run MALVA in haploid mode (default: false)

Positional arguments:
    <reference>     reference file in FASTA format (can be gzipped)
    <variants>      variants file in VCF format (can be gzipped)
    <sample>        sample file in FASTA/FASTQ format (can be gzipped)

Example

After you compiled malva, you can test it on the example data provided:

cd example
tar xvfz data.tar.gz
../MALVA -k 35 -r 43 -b 1 -f EUR_AF chr20.fa chr20.vcf chr20.sample.fa > chr20.genotyped.vcf

The last command is equivalent to run:

mkdir -p kmc_tmp
kmc -m4 -k43 -fm chr20.sample.fa kmc.out kmc_tmp
../bin/malva-geno index -k 35 -r 43 -b 1 -f EUR_AF chr20.fa chr20.vcf kmc.out
../bin/malva-geno call -k 35 -r 43 -b 1 -f EUR_AF chr20.fa chr20.vcf kmc.out > chr20.genotyped.vcf

This should take less than 1 minute to complete. You can also verify the correcteness of the output VCF chr20.genotyped.vcf by comparing it with chr20.malva.vcf. For a quick comparison look MALVA-TEST: Workflow for testing MALVA Output.

Haploid mode - Example

To run MALVA in haploid mode just use the -1 argument.

cd example
tar xvfz haploid.tar.gz
../MALVA -1 -k 35 -r 43 -b 1 -f AF haploid.fa haploid.vcf haploid.fq > haploid.genotyped.vcf

This should take less than 1 minute to complete. You can also verify the correcteness of the output VCF haploid.genotyped.vcf by comparing it with haploid.malva.vcf. For a quick comparison look MALVA-TEST: Workflow for testing MALVA Output.

Note

• The tool has been tested only on 64bit Linux system.

Authors

• Marco Previtali
• Luca Denti
• Giulia Bernardini
• Paola Bonizzoni
• Alexander Schönhuth
• Marco Burgio (code refactoring and malva-test utility)

For inquiries on this software please contact either MP or LD.

License

MALVA is distributed under the GPL-3.0-or-later license.