Hello VG team,

I want to compare alignment performances between Bowtie2, BWA-MEM2, and VG Giraffe. To do so, I have generated simulated reads (50 paired-end FASTQ files) based on chromosome 1 (HPRC v1.1) using vg sim and vg surject as follows:

vg sim \
    -x chr1.xg \
    -g chr1.gbwt \
    -n 20000000 \
    -p 500 -v 50 \
    -e 0.0024 -i 0.00029 \
    --threads 40 \
    --progress \
    -m HG02055 \
    -F REAL_SAMPLE_1.fastq.gz \
    -F REAL_SAMPLE_2.fastq.gz \
    --random-seed 42
    -a > HG02055.gam

vg surject \
    -x chr1.xg \
    --bam-output \
    --into-path CHM13#0#chr1 \
    --threads 40 \
    --progress \
    --sample HG02055 \
    HG02055.gam \
    | samtools reheader -c "sed s/CHM13#0#//g" - \
    | samtools sort -n -@ 40 - \
    | samtools fastq -@ 40 - > HG02055.interleaved.fastq

Testing the Bowtie2 and BWA-MEM2 is straightforward. However, when it comes to VG Giraffe, things get a bit complex. For instance, if I want to test sample HG02055, I should first generate a custom chr1 pangenome without paths associated with HG02055, as it would be similar to a personalized pangenome.

What is the best way to proceed from a computational perspective? I am aware I can remove a sample from a GWBT using the following command:

vg gbwt -o chr1.HG02055.gbwt --remove-sample HG02055 chr1.gbwt

However, how can I remove it from a GFA or GBZ graph?

Thank you in advance for any help you can provide.