Efficient deTection of CHromosomal rearrangements and fusIoN Genes

ETCHING takes about 3 hours for WGS data with 30X normal and 50X tumor on 30 threads on DELL 930 server. You can also find codes, k-mer set, and DEMO files on our website.

http://big.hanyang.ac.kr/ETCHING/download.html

We recommend using the new version of the pan-genome k-mer set, PGK2. You can download it from our website.

Using PGK2, ETCHING can detect somatic SVs from tumor sequencing data without matched-normal in comparable performance with other cutting-edge SV detection tools.

• Requirement changed
  • g++ >=7 (>=7.3 recommended)
  • cmake >=3.14
  • samtools >=1.13 (using htslib >=1.13)
• BAM mode accelerated
• CRAM support
• Minor bugs fixed

See CHANGE.md for older updates.

ETCHING was a bit slow in removing remaining germline SVs after scoring SVs. We solved this issue by increasing Samtools requirement to >=1.13 for collecting the paired-end reads relevant to the remaining artifacts. As we updated the step, we accelerated the workflow for BAM input along with CRAM support.

• 64-bit LINUX with >=32GB RAM (at least >=16GB).
• Tested on Fedora workstation, Centos, and Ubuntu

• Required to compile

  • gcc, g++ (>=7), make, cmake (>=3.14), wget

    Note: g++ >=7.3 recommended

• Required to run

  • BWA, samtools (>=1.13) using htslib (>=1.13)

We prepared a simple guide for CentOS/Fedora or Ubuntu/Debian/Mint users. You can skip this step if all requirements were installed.

Note: We tested this guide on Fedora32/33/34, CentOS7/8, Ubuntu16.04/18.04/20.04/22.04, Mint19/20, Debian11, and MX Linux.

• CentOS/Fedora (or other Red Hat-based Linux distros)

# Required programs 
sudo yum install -y gcc gcc-c++ make cmake bwa samtools wget

• Ubuntu/Debian/Mint (or other Debian-based distros)

## Required programs 
sudo apt install -y gcc g++ make cmake bwa samtools wget

Once the requirements were solved, you can install ETCHING as follows.

# Download ETCHING
git clone --depth=1 https://github.com/ETCHING-team/ETCHING.git

# Move to /path/to/ETCHING
cd ETCHING

# Installation (g++ >=7 and cmake >=3.14 required)
# g++ >=7.3 recommended
make

# Set your environment for ETCHING
echo "export PATH=$PATH:/path/to/ETCHING/bin" >> ~/.bashrc
exec $SHELL

# Change directory
cd /wherever/you/want/

# Download and decompress DEMO
wget http://big.hanyang.ac.kr/ETCHING/tiny_demo.tar.gz
# or you can find a demo file from our website
# http://big.hanyang.ac.kr/ETCHING/

# Decompress
tar zxvf tiny_demo.tar.gz
cd tiny_demo

# Run demo
etching -1 tumor_1.fq -2 tumor_2.fq -1c normal_1.fq -2c normal_2.fq -g Chr22.fa -f demo_PGK

You can run with bam or cram.

# or
etching -b tumor.sorted.cram -bc normal.sorted.cram -g Chr22.fa -f demo_PGK -p cram_mode_output

The Pan-Genome k-mer(PGK) set is used to build PGK filter. Since we updated the PGK to PGK2, you can also download the PGK2 as below. If you have no matched-normal data, you are highly recommended to use the PGK2 instead of PGK to call somatic SVs.

# Move to etching directory
mkdir -p /path/to
cd /path/to

# Download
wget http://big.hanyang.ac.kr/ETCHING/PGK2.tar.gz

# or you can download it from Amazon S3
# wget https://biglabhyu.s3.amazonaws.com/ETCHING/PGK.tar.gz

# Decompress
tar zxvf PGK2.tar.gz
cd PGK2

# Then, you must see PGK2.kmc_pre and PGK2_kmc_suf 
ls /path/to/PGK2/*
/path/to/PGK2/PGK2.kmc_pre
/path/to/PGK2/PGK2.kmc_suf

To use PGK2 (recommended), add -f /path/to/PGK2/PGK2 option. For instance,

etching -1 tumor_1.fq -2 tumor_2.fq -g reference.fa -f /path/to/PGK2/PGK2

Then, etching script will find /path/to/PGK2/PGK2.kmc_pre and /path/to/PGK2/PGK2.kmc_suf.

Alternatively, you can make your custom k-mer set using make_pgk included in ETCHING package.

In the absence of matched-normal sample, we recommend using >=200 genomes to build your custom k-mer set excluding rare k-mers (<1% allele frequency). If you have 600 genomes, use -m 6 option, which is the minimum frequency of k-mer to be included in your k-mer set.

make_pgk -i 600_genome.list -m 6 -o my_pgk

Here, 600_genome.list is a file of FASTA files of genomes in the format of

/path/to/references/hg19.fasta
/path/to/references/GRCh38.fa
/path/to/references/NA12878.chr1.fna
/path/to/references/NA12878.chr2.fna
.
.
.
/path/to/assemblies/someones_genome.scaffold_1.fa
/path/to/assemblies/someones_genome.scaffold_2.fa

docker

# Download ETCHING docker image
wget http://big.hanyang.ac.kr/ETCHING/ETCHING_v1.4.2.docker.tar

# Load the image
docker load -i ETCHING_v1.4.2.docker.tar

# Check the image
docker images

You can see like below (numbers can be different)

Download demo data and run ETCHING with docker

wget http://big.hanyang.ac.kr/ETCHING/tiny_demo.tar.gz

tar zxvf tiny_demo.tar.gz

docker run -i -t --rm -v $PWD/tiny_demo/:/work/ etching:1.4.2 etching \
-1 tumor_1.fq -2 tumor_2.fq -1c normal_1.fq -2c normal_2.fq -g Chr22.fa -f demo_PGK -t 8

Here, etching:1.4.2 is REPOSITORY:TAG of ETCHING docker image.

To use your data, replace $PWD/tiny_demo with /your/data/path/. The directory also should contain k-mer set (PGK2).

Note: Keep /work/ in the above command line, since it's the default working directory.

An older version of ETCHING (v1.4.0) is also available on Amazon Web Service (AMI ID: ami-07c7a7d8934784df9; Region: us-east-1 (Northern Virginia)).

# Lunch EC2 instance from ETCHING AMI
# And connect to EC2 instance
ssh -i Your_Key ubuntu@Your_Instance_Address

# Decompress
tar zxvf ~/resources/DEMO.tar.gz 

# Run demo
cd ~/resources/DEMO
bash example.sh

https://github.com/ETCHING-team/LR_Filter

https://github.com/ETCHING-team/etching_bench

Sohn, Ji., Choi, MH., Yi, D. et al. Ultrafast prediction of somatic structural variations by filtering out reads matched to pan-genome k-mer sets. Nat. Biomed. Eng (2022). https://doi.org/10.1038/s41551-022-00980-5

Jang-il Sohn, Min-Hak Choi, Dohun Yi, A. Vipin Menon, and Jin-Wu Nam

Bioinformatics and Genomics Lab., Dept. of Life Science, Hanyang University, Seoul 04763, Korea

Jin-Wu Nam ([email protected])

Jang-il Sohn ([email protected])

Min-Hak Choi ([email protected])

Dohun Yi ([email protected])