📊 Long-read SV Benchmark CMRG

This is a reproducible benchmark of structural variants in challenging medically relevant genes. The truth set is described in detail here:

Curated variation benchmarks for challenging medically relevant autosomal genes. Wagner et al., 2022. Nature Biotechnology

	caller	platform	TP	FP	FN	precision	recall	f1	gt_concordance
0	sniffles	ONT	194	18	23	0.9151	0.894	0.9044	0.8814
1	cuteSV	ONT	194	13	23	0.9372	0.894	0.9151	0.8918
2	delly	ONT	193	12	24	0.9415	0.8894	0.9147	0.8964
3	dysgu	ONT	195	1	22	0.9949	0.8986	0.9443	0.8821

	caller	platform	TP	FP	FN	precision	recall	f1	gt_concordance
0	sniffles	PacBio	193	4	24	0.9797	0.8894	0.9324	0.8912
1	cuteSV	PacBio	190	8	27	0.9596	0.8756	0.9157	0.9
2	delly	PacBio	191	5	26	0.9745	0.8802	0.9249	0.8743
3	dysgu	PacBio	193	1	24	0.9948	0.8894	0.9392	0.8808

Reads were from Oxford Nanopore kit14 (~40X coverage), and PacBio Revio HiFi (~30X coverage). SV callers tested were as follows:

• sniffles v2.2.0

• cuteSV v2.0.3

• delly v1.1.6

• dysgu v1.6.0

For benchmarking truvari v4.0.0 was used with parameters -r 1000 --passonly

Run the benchmark.sh script or follow along below.

Requirements:

• ~ 200 Gb space, 32 gb Ram, 4 cores
• Docker / Singularity (Required for Mac, optional for Linux)

Setup environment

mkdir benchmark && cd benchmark
docker run -it --memory="32g" --mount src="${PWD}",target=/results,type=bind condaforge/mambaforge
mamba update conda -y && cd results

Note, you may need to set the memory and swap space manually using Docker Desktop on Mac.

Install tools:

mamba create -c bioconda -c conda-forge -n bench python=3.9 awscli sniffles=2.2.0 cuteSV=2.0.3 truvari=4.0.0 delly=1.1.6 -y
conda activate bench
pip install dysgu==1.6.0

Grab datasets

Reference genome:

wget https://ftp.ncbi.nlm.nih.gov/genomes/all/GCA/000/001/405/GCA_000001405.15_GRCh38/seqs_for_alignment_pipelines.ucsc_ids/GCA_000001405.15_GRCh38_no_alt_analysis_set.fna.gz
wget https://ftp.ncbi.nlm.nih.gov/genomes/all/GCA/000/001/405/GCA_000001405.15_GRCh38/seqs_for_alignment_pipelines.ucsc_ids/GCA_000001405.15_GRCh38_no_alt_analysis_set.fna.fai
gunzip GCA_000001405.15_GRCh38_no_alt_analysis_set.fna.gz
ref=GCA_000001405.15_GRCh38_no_alt_analysis_set.fna

Oxford Nanopore reads https://labs.epi2me.io/giab-2023.05/:

aws s3 sync --no-sign-request --include='PAO89685.pass.cram*' --exclude="*fail*" s3://ont-open-data/giab_2023.05/analysis/hg002/sup/ .

PacBio reads https://www.pacb.com/connect/datasets/:

wget https://downloads.pacbcloud.com/public/revio/2022Q4/HG002-rep1/analysis/HG002.m84011_220902_175841_s1.GRCh38.bam
wget https://downloads.pacbcloud.com/public/revio/2022Q4/HG002-rep1/analysis/HG002.m84011_220902_175841_s1.GRCh38.bam.bai

SV truth set:

r=ftp://ftp-trace.ncbi.nlm.nih.gov/giab/ftp/data/AshkenazimTrio/analysis/NIST_HG002_medical_genes_SV_benchmark_v0.01/HG002_GRCh38_difficult_medical_gene_SV_benchmark_v0.01
wget ${r}.bed && wget ${r}.vcf.gz && wget ${r}.vcf.gz.tbi

Run SV callers

ONT reads:

sniffles --input PAO89685.pass.cram --vcf HG002.PAO89685.sniffles.vcf

dysgu run --mode nanopore --procs 4 -x --clean $ref wd PAO89685.pass.cram > HG002.PAO89685.dysgu.vcf

mkdir wd_cuteSV
cuteSV -t 4 -s 3 --genotype PAO89685.pass.cram $ref HG002.PAO89685.cuteSV.vcf wd_cuteSV

delly lr -g $ref PAO89685.pass.cram > HG002.PAO89685.delly.vcf

PacBio reads:

sniffles --input HG002.m84011_220902_175841_s1.GRCh38.bam --vcf HG002.pacbio.sniffles.vcf

dysgu call --mode pacbio --procs 4 -x --clean $ref wd HG002.m84011_220902_175841_s1.GRCh38.bam > HG002.pacbio.dysgu.vcf

mkdir wd_cuteSV
cuteSV -t 4 -s 3 --genotype HG002.m84011_220902_175841_s1.GRCh38.bam $ref HG002.pacbio.cuteSV.vcf wd_cuteSV

delly lr -g $ref  HG002.m84011_220902_175841_s1.GRCh38.bam > HG002.pacbio.delly.vcf

Benchmark

ONT data:

callers=( "sniffles" "cuteSV" "delly" "dysgu" )

for name in "${callers[@]}"
do
  bgzip HG002.PAO89685.${name}.vcf
  tabix HG002.PAO89685.${name}.vcf.gz
  truvari bench -f GCA_000001405.15_GRCh38_no_alt_analysis_set.fna \
                -b HG002_GRCh38_difficult_medical_gene_SV_benchmark_v0.01.vcf.gz \
                --includebed HG002_GRCh38_difficult_medical_gene_SV_benchmark_v0.01.bed \
                -c HG002.PAO89685.${name}.vcf.gz \
                --passonly -r 1000 \
                -o truvari_${name}
done

PacBio data:

callers=( "sniffles" "cuteSV" "delly" "dysgu" )

for name in "${callers[@]}"
do
  bgzip HG002.pacbio.${name}.vcf
  tabix HG002.pacbio.${name}.vcf.gz
  truvari bench -f GCA_000001405.15_GRCh38_no_alt_analysis_set.fna \
                -b HG002_GRCh38_difficult_medical_gene_SV_benchmark_v0.01.vcf.gz \
                --includebed HG002_GRCh38_difficult_medical_gene_SV_benchmark_v0.01.bed \
                -c HG002.pacbio.${name}.vcf.gz \
                --passonly -r 1000 \
                -o truvari_${name}
done

Run the plotting script:

python3 plot_benchmark.py