Decoil

Decoil (deconvolve extrachromosomal circular DNA isoforms from long-read data) is a software package for reconstruction circular DNA.

• Getting started using conda and pip
• Getting started using docker
• Getting started using singularity
• Test example for docker or singularity
• Install Decoil from source (latest features, unstable)
• Decoil run configuration
• File formats
• FAQ
• Citation
• License

Getting started using conda and pip

Assumes you have conda installed.

# install conda dependencies
CONDAENV="envdecoil"
# linux
conda create -n $CONDAENV --override-channels -c bioconda -c conda-forge python==3.9 survivor==1.0.7 sniffles==1.0.12 ngmlr==0.2.7 samtools datrie
# macos
conda create -n $CONDAENV --override-channels -c bioconda -c conda-forge python==3.9 survivor==1.0.7 sniffles==1.0.7 ngmlr==0.2.7 samtools datrie --platform osx-64

conda activate $CONDAENV

# install decoil via pip
python -m pip install decoil==2.0.2

# optional
export PATH=~/miniconda3/envs/$CONDAENV/bin:$PATH

# check if decoil in path
which decoil
# check version
decoil --version

Getting started using docker

As a prerequisite you need to have installed docker (you can install this from the official website or using conda).

Download as docker image

Download decoil docker image from docker-hub. This contains all the dependencies needed to run the software. No additional installation needed. All the environment, packages, dependencies are all specified in the docker/singularity image.

# docker
docker pull madagiurgiu25/decoil:2.0.2

Run example using docker (optional)

Test docker installation using example.

Run Decoil reconstruction using docker

To run Decoil on your data you need to configure the following parameters:

# run decoil with your input with standard parameters
BAM_INPUT="<absolute path to your BAM file>"
OUTPUT_FOLDER="<absolute path to your output folder>"
NAME="<sample name>"
GENOME="<absolute path to your reference genome file>"
ANNO="<absolute path to your gtf annotation file>"

and then run the following command:

# docker
docker run -it --platform=linux/amd64 \
    -v ${BAM_INPUT}:/data/input.bam \
    -v ${BAM_INPUT}.bai:/data/input.bam.bai \
    -v ${GENOME}:/annotation/reference.fa \
    -v ${ANNO}:/annotation/anno.gtf \
    -v ${OUTPUT_FOLDER}:/mnt \
    -t madagiurgiu25/decoil:2.0.2 \
    decoil-pipeline sv-reconstruct \
            -b /data/input.bam \
            -r /annotation/reference.fa \
            -g /annotation/anno.gtf \
            -o /mnt --name ${NAME}

To test your installation using example.

Getting started using singularity

As a prerequisite you need to have installed singularity (you can install this from the official website or using conda).

Download as singularity image

# singularity
singularity pull decoil.sif  docker://madagiurgiu25/decoil:2.0.2

Run example using singularity (optional)

Test singularity installation using example.

Run Decoil reconstruction using singularity

To run Decoil on your data you need to configure the following parameters:

# run decoil with your input with standard parameters
BAM_INPUT="<absolute path to your BAM file>"
OUTPUT_FOLDER="<absolute path to your output folder>"
NAME="<sample name>"
GENOME="<absolute path to your reference genome file>"
ANNO="<absolute path to your gtf annotation file>"

and then run the following command:

# singularity
mkdir -p ${OUTPUT_FOLDER}
mkdir -p ${OUTPUT_FOLDER}/logs
mkdir -p ${OUTPUT_FOLDER}/tmp
singularity run \
    --bind ${OUTPUT_FOLDER}/logs:/mnt/logs \
    --bind ${OUTPUT_FOLDER}/tmp:/tmp \
    --bind ${BAM_INPUT}:/data/input.bam \
    --bind ${BAM_INPUT}.bai:/data/input.bam.bai \
    --bind ${GENOME}:/annotation/reference.fa \
    --bind ${ANNO}:/annotation/anno.gtf \
    --bind ${OUTPUT_FOLDER}:/mnt \
    decoil.sif \
    decoil-pipeline sv-reconstruct \
            -b /data/input.bam \
            -r /annotation/reference.fa \
            -g /annotation/anno.gtf \
            -o /mnt --name ${NAME}

Test example for docker or singularity

To test docker and singularity installation use the example.

Install Decoil from source (latest features, unstable)

You can install the latest version of Decoil repository. Note this is an unstable version and contains bugs. git and conda/mamba are prerequisites.

Linux

# create conda environment
conda create -n envdecoil -c bioconda -c conda-forge python==3.10 survivor==1.0.7 sniffles==1.0.12 ngmlr==0.2.7 samtools==1.15.1 deeptools==3.5.5
conda activate envdecoil

# install decoil
git clone https://github.com/madagiurgiu25/decoil-pre.git
cd decoil-pre
python -m pip install -r requirements.txt
python setup.py install

And check if the installation worked:

# might take a while
decoil-pipeline --version
decoil --version

MacOS

# create conda environment
conda create -n envdecoil -c bioconda -c conda-forge python==3.10 survivor==1.0.7 sniffles==1.0.7 ngmlr==0.2.7 samtools==1.15.1 --platform osx-64
conda activate envdecoil

# install decoil
git clone https://github.com/madagiurgiu25/decoil-pre.git
cd decoil-pre
python -m pip install -r requirements.txt
python setup.py install

And check if the installation worked:

# might take a while
decoil-pipeline --version
decoil --version

Decoil run configurations

An overview about the available functionalities:

	decoil-pipeline	decoil	decoil-viz
	(recommended)	(advanced users)	(recommended)
SV calling	x
coverage track	x
reconstruction	x	x
visualization			x
docker	x	x	x
singularity	x	x	x

1. Reconstruct ecDNA using decoil-pipeline (recommended)

To reconstruct ecDNA we recommend to use decoil-pipeline using the sv-reconstruct mode.
This requires only a .bam file as input and generates internally all the files required for the reconstruction.

# call help
docker run -it --platform=linux/amd64 -t madagiurgiu25/decoil:2.0.2 decoil-pipeline --help

usage: decoil-pipeline <workflow> <parameters> [<target>]
Example: 
    # run decoil including the processing and visualization steps
    decoil-pipeline -f sv-recontruct --bam <input> --outputdir <outputdir> --name <sample> --sv-caller <sniffles> -r <reference-genome> -g <annotation-gtf>
        

Decoil 1.1.2: reconstruct ecDNA from long-read data

positional arguments:
  {sv-only,sv-reconstruct,reconstruct-only}
                        sub-command help
    sv-only             Perform preprocessing
    sv-reconstruct      Perform preprocessing and reconstruction

optional arguments:
  -h, --help            show this help message and exit
  --version             show program's version number and exit
  -n, --dry-run
  -f, --force
  -c, --use-conda

You can run decoil-pipeline using following modes:

• sv-only
• sv-reconstruct
• reconstruct-only

Check the description in running modes.

2. Reconstruct ecDNA using decoil (advanced users only)

This configuration is the most flexible and allows users to use their own SV calls. For details go here.

3. Visualization of ecDNA threads using decoil-viz (recommended)

To interpret and visualize the results of the ecDNA reconstruction threads, use decoil-viz.

FAQ

Check recommendations for filtering or debugging in the FAQ section.

File formats

The relevant output files for the users are:

• reconstruct.bed - contains all genomic fragments in order composing for all reconstructions
• reconstruct.ecDNA.bed - contains all genomic fragments in order composing reconstructions labeled as ecDNA
• reconstruct.ecDNA.filtered.bed - contains all genomic fragments in order composing the reconstructios labeled as ecDNA and passing the --filter-score
• summary.txt - summarize all the circular reconstructions

Example reconstruct.bed:

cat reconstruct.bed

#chr    start   end     circ_id fragment_id     strand  coverage        estimated_proportions
chr2    15585356        15633376        0       5       +       149     75
chr3    11150000        11160001        0       41      -       103     75
chr3    11049997        11060001        0       33      +       117     75
chr2    15585356        15633376        3       5       +       149     36
chr3    11150000        11160001        3       41      -       103     36
chr3    11049997        11060001        3       33      +       117     36
chr2    15585356        15633376        3       5       +       149     36
chr2    16521052        16628305        3       13      +       37      36
chr3    10981202        11028470        3       25      -       31      36
chr12   68807722        68970910        2       53      +       252     252

Column	Description
chr	Chromosome containing the genomic fragment.
start	Start coordinate of the fragment.
end	End coordinate of the fragment.
circ_id	Identifier of the reconstructed circular DNA molecule. Fragments with the same circ_id belong to the same reconstruction.
fragment_id	Unique identifier of the genomic fragment used in the reconstruction.
strand	Orientation (+ or -) of the fragment within the reconstructed cycle.
coverage	Sequencing coverage (read depth) supporting this genomic fragment.
estimated_proportions	Estimated abundance of the reconstructed cycle. This value is identical for all fragments belonging to the same circ_id.

Example summary.txt:

cat summary.txt

circ_id chr_origin      size(MB)        label   topology_idx    topology_name   estimated_proportions
0       chr3,chr2       0.068025                4       multi_region_inter_chr  75
3       chr3,chr2       0.270566        ecDNA   5       simple_duplications     36
2       chr12           0.163188        ecDNA   0       simple_circle           252

Column	Description
circ_id	Identifier of predicted cycle.
chr_origin	Chromosome(s) contributing fragments to the reconstructed cycle. Multiple chromosomes are comma-separated.
size(MB)	Total size of the reconstructed cycle in megabases (Mb).
label	Classification assigned to the reconstruction (e.g. ecDNA). May be empty if no label is assigned.
topology_idx	Numeric identifier of the inferred structural topology.
topology_name	Human-readable name of the inferred topology (e.g. simple_circle, multi_region_inter_chr, simple_duplications).
estimated_proportions	Estimated abundance of the reconstructed structure.

Citation

If you use Decoil for your work please cite our paper:

Madalina Giurgiu, Nadine Wittstruck, Elias Rodriguez-Fos, Rocio Chamorro Gonzalez, Lotte Bruckner, Annabell Krienelke-Szymansky, Konstantin Helmsauer, Anne Hartebrodt, Philipp Euskirchen, Richard P. Koche, Kerstin Haase*, Knut Reinert*, Anton G. Henssen*. Reconstructing extrachromosomal DNA structural heterogeneity from long-read sequencing data using Decoil. Genome Research 2024, DOI: https://doi.org/10.1101/gr.279123.124

@article{Giurgiu2024ReconstructingDecoil,
    title = {{Reconstructing extrachromosomal DNA structural heterogeneity from long-read sequencing data using Decoil}},
    year = {2024},
    journal = {Genome Research},
    author = {Giurgiu, Madalina and Wittstruck, Nadine and Rodriguez-Fos, Elias and Chamorro Gonzalez, Rocio and Brueckner, Lotte and Krienelke-Szymansky, Annabell and Helmsauer, Konstantin and Hartebrodt, Anne and Euskirchen, Philipp and Koche, Richard P. and Haase, Kerstin and Reinert, Knut and Henssen, Anton G.},
    month = {8},
    pages = {gr.279123.124},
    doi = {10.1101/gr.279123.124},
    issn = {1088-9051}
}

Paper repository: https://github.com/henssen-lab/decoil-paper

License

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

Disclaimer

Decoil and the content of this research-repository (i) is not suitable for a medical device; and (ii) is not intended for clinical use of any kind, including but not limited to diagnosis or prognosis.