Delly

Delly is an integrated structural variant (SV) prediction method that can discover, genotype and visualize deletions, tandem duplications, inversions and translocations at single-nucleotide resolution in short-read and long-read massively parallel sequencing data. It uses paired-ends, split-reads and read-depth to sensitively and accurately delineate genomic rearrangements throughout the genome.

Installing Delly

Delly is available as a statically linked binary, a singularity container (SIF file), a docker container or via Bioconda. You can also build Delly from source using a recursive clone and make.

git clone --recursive https://github.com/dellytools/delly.git

cd delly/

make all

There is a Delly discussion group delly-users for usage and installation questions.

Delly multi-threading mode

Delly supports parallel computing using the OpenMP API (www.openmp.org).

make PARALLEL=1 src/delly

You can set the number of threads using the environment variable OMP_NUM_THREADS.

export OMP_NUM_THREADS=2

Delly primarily parallelizes on the sample level. Hence, OMP_NUM_THREADS should be always smaller or equal to the number of input samples.

Running Delly

Delly needs a sorted, indexed and duplicate marked bam file for every input sample. An indexed reference genome is required to identify split-reads. Common workflows for germline and somatic SV calling are outlined below.

delly call -g hg38.fa input.bam > delly.vcf

You can also specify an output file in BCF format.

delly call -o delly.bcf -g hg38.fa input.bam

bcftools view delly.bcf > delly.vcf

Example

A small example is included for short-read, long-read and copy-number variant calling.

delly call -g example/ref.fa -o sr.bcf example/sr.bam

delly lr -g example/ref.fa -o lr.bcf example/lr.bam

delly cnv -g example/ref.fa -m example/map.fa.gz -c out.cov.gz -o cnv.bcf example/sr.bam

More in-depth tutorials for SV calling are available here:

• Short-read SV calling: https://github.com/tobiasrausch/vc

• Long-read SV calling: https://github.com/tobiasrausch/sv

Somatic SV calling

• At least one tumor sample and a matched control sample are required for SV discovery

delly call -x hg38.excl -o t1.bcf -g hg38.fa tumor1.bam control1.bam

• Somatic pre-filtering requires a tab-delimited sample description file where the first column is the sample id (as in the VCF/BCF file) and the second column is either tumor or control.

delly filter -f somatic -o t1.pre.bcf -s samples.tsv t1.bcf

• Genotype pre-filtered somatic sites across a larger panel of control samples to efficiently filter false postives and germline SVs. For performance reasons, this can be run in parallel for each sample of the control panel and you may want to combine multiple pre-filtered somatic site lists from multiple tumor samples.

delly call -g hg38.fa -v t1.pre.bcf -o geno.bcf -x hg38.excl tumor1.bam control1.bam ... controlN.bam

• Post-filter for somatic SVs using all control samples.

delly filter -f somatic -o t1.somatic.bcf -s samples.tsv geno.bcf

Germline SV calling

• SV calling is done by sample for high-coverage genomes or in small batches for low-coverage genomes

delly call -g hg38.fa -o s1.bcf -x hg38.excl sample1.bam

• Merge SV sites into a unified site list

delly merge -o sites.bcf s1.bcf s2.bcf ... sN.bcf

• Genotype this merged SV site list across all samples. This can be run in parallel for each sample.

delly call -g hg38.fa -v sites.bcf -o s1.geno.bcf -x hg38.excl s1.bam

delly call -g hg38.fa -v sites.bcf -o sN.geno.bcf -x hg38.excl sN.bam

• Merge all genotyped samples to get a single VCF/BCF using bcftools merge

bcftools merge -m id -O b -o merged.bcf s1.geno.bcf s2.geno.bcf ... sN.geno.bcf

• Apply the germline SV filter which requires at least 20 unrelated samples

delly filter -f germline -o germline.bcf merged.bcf

Delly for long reads from PacBio or ONT

Delly also supports long-reads for SV discovery.

delly lr -y ont -o delly.bcf -g hg38.fa input.bam

delly lr -y pb -o delly.bcf -g hg38.fa input.bam

Alternate alignments for genome graphs

Instead of providing only one input alignment, delly supports now multiple alternate alignments on different linear reference genomes using minimap2 or pan-genome graphs using minigraph.

minimap2 -ax map-pb -L chm13.fa sample.fq.gz
minigraph --vc -cx lr pangenome.gfa.gz sample.fq.gz

If the above alignment files are then stored as sample.chm13.bam and sample.gaf.gz you can use a simple tab-delimited config file for all alternate alignments with delly.

cat align.config

sample.chm13.bam   chm13.fa
sample.gaf.gz   pangenome.gfa.gz

delly lr -y pb -o delly.bcf -g hg38.fa -l align.config sample.hg38.bam

Structural variants are still reported with respect to GRCh38 coordinates but the output will only contain SVs that are not present in any of the alternate alignments. For pangenome graphs you may want to try the augmented graph from this study. Please note that this graph contains only SVs greater 50bp so you need to filter the above delly output to match the size range using bcftools.

bcftools view -i '(QUAL>=300) && ( ((SVTYPE=="INS") && (INFO/SVLEN>50)) || (SVTYPE="BND") || ((INFO/END - POS)>50) )' delly.bcf

Read-depth profiles and copy-number variant calling

You can generate read-depth profiles with delly. This requires a mappability map which can be downloaded here:

Mappability Maps

The command to count reads in 10kbp mappable windows and normalize the coverage is:

delly cnv -a -g hg38.fa -m hg38.map -c out.cov.gz -o out.bcf input.bam

The output file out.cov.gz can be plotted using R to generate normalized copy-number profiles and segment the read-depth information:

Rscript R/rd.R out.cov.gz

Instead of segmenting the read-depth information, you can also visualize the CNV calls.

bcftools query -f "%CHROM\t%POS\t%INFO/END\t%ID[\t%RDCN]\n" out.bcf > seg.bed

Rscript R/rd.R out.cov.gz seg.bed

With -s you can output a statistics file with GC bias information.

delly cnv -g hg38.fa -m hg38.map -c out.cov.gz -o out.bcf -s stats.gz input.bam

zcat stats.gz | grep "^GC" > gc.bias.tsv

Rscript R/gcbias.R gc.bias.tsv

Germline CNV calling

Delly uses GC and mappability fragment correction to call CNVs. This requires a mappability map.

• Call CNVs for each sample and optionally refine breakpoints using delly SV calls

delly cnv -o c1.bcf -g hg38.fa -m hg38.map -l delly.sv.bcf input.bam

• Merge CNVs into a unified site list

delly merge -e -p -o sites.bcf -m 1000 -n 100000 c1.bcf c2.bcf ... cN.bcf

• Genotype CNVs for each sample

delly cnv -u -v sites.bcf -g hg38.fa -m hg38.map -o geno1.bcf input.bam

• Merge genotypes using bcftools

bcftools merge -m id -O b -o merged.bcf geno1.bcf ... genoN.bcf

• Filter for germline CNVs

delly classify -f germline -o filtered.bcf merged.bcf

• Optional: Plot copy-number distribution for large number of samples (>>100)

bcftools query -f "%ID[\t%RDCN]\n" filtered.bcf > plot.tsv

Rscript R/cnv.R plot.tsv

Somatic copy-number alterations (SCNAs)

• For somatic copy-number alterations, delly first segments the tumor genome (-u is required). Depending on the coverage, tumor purity and heterogeneity you can adapt parameters -z, -t and -x which control the sensitivity of SCNA detection.

delly cnv -u -z 10000 -o tumor.bcf -c tumor.cov.gz -g hg38.fa -m hg38.map tumor.bam

• Then these tumor SCNAs are genotyped in the control sample (-u is required).

delly cnv -u -v tumor.bcf -o control.bcf -g hg38.fa -m hg38.map control.bam

• The VCF IDs are matched between tumor and control. Thus, you can merge both files using bcftools.

bcftools merge -m id -O b -o tumor_control.bcf tumor.bcf control.bcf

• Somatic filtering requires a tab-delimited sample description file where the first column is the sample id (as in the VCF/BCF file) and the second column is either tumor or control.

delly classify -p -f somatic -o somatic.bcf -s samples.tsv tumor_control.bcf

• Optional: Plot the SCNAs using bcftools and R.

bcftools query -s tumor -f "%CHROM\t%POS\t%INFO/END\t%ID[\t%RDCN]\n" somatic.bcf > segmentation.bed

Rscript R/rd.R tumor.cov.gz segmentation.bed

FAQ

• Visualization of SVs
  You may want to try out wally to plot candidate structural variants. The paired-end coloring is explained in wally's README file.

• What is the smallest SV size Delly can call?
  For short-reads, this depends on the sharpness of the insert size distribution. For an insert size of 200-300bp with a 20-30bp standard deviation, Delly starts to call reliable SVs >=300bp. Delly also supports calling of small InDels using soft-clipped reads only, the smallest SV size called is 15bp. For long-reads, delly calls SVs >=30bp.

• Can Delly be used on a non-diploid genome?
  Yes and no. The SV site discovery works for any ploidy. However, Delly's genotyping model assumes diploidy (hom. reference, het. and hom. alternative). The CNV calling allows to set the baseline ploidy on the command-line.

• Delly is running too slowly what can I do?
  You should exclude telomere and centromere regions and also all unplaced contigs (-x command-line option). In addition, you can filter input reads more stringently using -q 20 and -s 15. Lastly, -z can be set to 5 for high-coverage data.

• Are non-unique alignments, multi-mappings and/or multiple split-read alignments allowed?
  Delly expects two alignment records in the bam file for every paired-end, one for the first and one for the second read. Multiple split-read alignment records of a given read are allowed if and only if one of them is a primary alignment whereas all others are marked as secondary or supplementary. This is the default for bwa, minimap2 and many other aligners.

• What pre-processing of bam files is required?
  Bam files need to be sorted, indexed and ideally duplicate marked.

• Usage/discussion mailing list?
  There is a delly discussion group delly-users.

• Docker/Singularity support?
  There is a delly docker container and singularity container (*.sif file) available.

• How can I compute a mappability map?
  A basic mappability map can be built using dicey, samtools and bwa with the below commands (as an example for the sacCer3 reference):

dicey chop sacCer3.fa
bwa index sacCer3.fa
bwa mem sacCer3.fa read1.fq.gz read2.fq.gz | samtools sort -@ 8 -o srt.bam -
samtools index srt.bam 
dicey mappability2 srt.bam 
gunzip map.fa.gz && bgzip map.fa && samtools faidx map.fa.gz 

• Bioconda support?
  Delly is available via bioconda.

Citation

Tobias Rausch, Thomas Zichner, Andreas Schlattl, Adrian M. Stuetz, Vladimir Benes, Jan O. Korbel.
DELLY: structural variant discovery by integrated paired-end and split-read analysis.
Bioinformatics. 2012 Sep 15;28(18):i333-i339.
https://doi.org/10.1093/bioinformatics/bts378

License

Delly is distributed under the BSD 3-Clause license. Consult the accompanying LICENSE file for more details.

Credits

HTSlib is heavily used for all genomic alignment and variant processing. Boost for various data structures and algorithms and Edlib for pairwise alignments using edit distance.