Please download test data from zenodo (opens in a new tab) and extract files with tar xf test_data.tar.gz. The directory follows the following structure:

hprc.vcf.gz contains a subset of pangenomic VCF file (see here). Target loci are described by a four-column file loci.bed.

In this example, we have GRCh38 reference genome at genome.fa. Such reference genome can be downloaded here (opens in a new tab) (Genome sequence, primary assembly, direct link (opens in a new tab)). Please index the genome and construct k-mer counts (instructions):

samtools faidx genome.fa
jellyfish count --canonical --lower-count 2 --out-counter-len 2 --mer-len 25 \
    --threads 8 --size 3G --output counts.jf genome.fa

Then, you can create the database of loci using

locityper add -d db -v hprc.vcf.gz -r genome.fa -j counts.jf -L loci.bed

Most importantly, this command will produce locus haplotypes:

Next, please run WGS dataset preprocessing:

locityper preproc -i reads/{1,2}.fastq.gz -r genome.fa -j counts.jf -o bg

This command will align a fraction of the reads to a part of chromosome 17 in order to identify background read depth distribution and error profiles.

Finally, you can run targeted genotyping with

locityper genotype -i reads/{1,2}.fastq.gz -d db -p bg -o gts

This will produce genotype predictions:

You can find output file descriptions here. In addition, you can summarize JSON files in one CSV file with

/path/to/locityper/extra/into_csv.py -i ./gts -o gts.csv

Note: into_csv takes the first segment after . as sample name, so sample name in the output CSV file will be gts.

Due to non-determenistic steps within Locityper, results may differ. Sample CSV file may look like this:

sample  locus  genotype             quality  total_reads  unexpl_reads  weight_dist  warnings
gts     MUC6   HG00621.1,HG00621.2  108.8    3738         1             0.00000      *
gts     MUC16  HG00621.1,HG00621.2  585.8    13969        16            0.00000      *

In total, all steps should take 2–4 minutes.