During cell death the DNA of tumor cells is released in to the bloodstream, forming a pool of circulating cell-free DNA (cfDNA). cfDNA can be isolated from blood and/or urine of cancer patients and analyzed with sequencing. Most conventional short-read sequencing methods are technically challenging, labor intensive and time consuming, requiring several days but more typically weeks to obtain interpretable data which are limited by a bias for short cfDNA fragments. The ITSFASTR toolkit provides the bioinformatic workflow for the analysis of copy number aberrations, fragmentation and fragment end sequences and nucleosome positioning from conventional Illumina and Oxford Nanopore Technologies (ONT) cfDNA data.
The ITSFASTR pipeline is designed to run on HPC clusters, as some steps can be resource-heavy with real-world data sets.
If ran on a standard computer, we recommend: RAM: 16+ GB CPU: 4+ cores, 3.3+ GHz/core
The package is tested on an Ubuntu 20.04.4 LTS operating system, but it should work on Windows Subsystem for Linux (WSL) as well.
Prior to downloading ITSFASTR users should install Anaconda.
ITSFASTR uses Snakemake for reproducibility. Install Snakemake by running:
conda create -n snakemake -c bioconda snakemakeActivate your environment:
conda activate snakemakeSnakemake uses Mamba, a fast package manager. To install it type:
conda install mamba -n base -c conda-forgeSnakemake requires strict channel priorities in Conda, which is also good for reproducibility. To set this, type:
conda config --set channel_priority strictEvery other dependency will be installed automatically during the first run of the pipeline. Software versions used by the pipeline can be found in the environment files in the envs folder.
In short the pipeline uses: BBduk or Cutadapt for adapter trimming of Illumina reads or Porechop for ONT reads. BWA MEM for mapping Illumina and Minimap2 for ONT reads. Samtools and Sambamba for quality filtering and marking duplicates. Qualimap for quality control summarized by MultiQC.
Download ITSFASTR by:
wget https://github.com/mouliere-lab/ITSFASTR/archive/refs/heads/main.zip &&
unzip main.zip &&
mv ITSFASTR-main ITSFASTR &&
rm main.zipor if you have Git installed:
git clone https://github.com/mouliere-lab/ITSFASTR.gitThe ITSFASTR pipeline accepts paired Illumina sequencing files with
.fq.gz extension. If your files have a different extension use the
script in workflow/scripts/0_raw_seq/1_raw_renaming.sh to change it.\
ONT sequencing files can be fed into the pipeline after demultiplexing.
For an ONT run a folder containing the sequencing files is the input,
for ex. each barcode folder. For further explanation see the structure of
the provided dummy data.
To run this demo please download these dummy files. The dummy files are a synthetic admixtures of real Illumina and ONT sequencing files from the plasma of cancer patients, downsampled to 100K reads.
The sample sheet is a space-delimited .csv file
with two labels in its header: sample_name group
An example containing the dummy file names can be found in the config directory.
sample_name is the unique name of the sample (file). Paired input files
have one unique sample name, which is the file name minus
the _R1(2).fq.gz suffix.
group represents the single analysis group the given file belongs to. For the
final steps of the FrEIA analysis a control group is needed. Mark this with a
* following the group name.
The config file is a .yaml format file containing the
parameters used during the run.
This will be used as the name of the main folder containing results.
Based on the sequencing files either select Illumina or ONT.
For an Illumina run add the path to the folder containing the .fq.gz files.\
For an ONT run add the path to the folder containing the output folder(s) of
base calling or demultiplexing.
Add the path to the sample sheet file.
The pipeline maps the reads to a reference genome using either BWA MEM for Illumina runs or Minimap2 for ONT runs. To map the reads you first need to index the reference using BWA MEM (tutorial) for Illumina runs or Minimap2 (tutorial) for ONT runs. Indexing large genomes can take from minutes to hours butt needs to be done only once. Add the path to the indexed reference genome based on the run type.
Add the path to the output folder. A folder with the project name will be created here containing all the results.
The temporary folder (tmp) will be used to store temporary files. On HPC clusters
this should be on a scratch drive with fast read/write capability.
The number of CPU cores should be less or equal to the available cores. On a normal computer this is usually 4 or 8. Be aware that with the increase of the core number the memory usage can also increase.
The mappability track used by Griffin LR is big, thus it is not provided. Download a mappability track and place it in /resources/Griffin_LR.
Set the path to the reference genome fasta file.
Griffin LR requires one or more sites of interest file. Each file can contain multiple sites of the same tipe. For an example please see the Running ITSFASTR on a local machine using Snakemake.
This is the recommended way of running the pipeline, as some steps are
resource-heavy.
Rules of the pipeline are run separately. To run a rule, first change directories
into the rule's folder for example cd workflow/rules/1_preprocessing/
and run snakemake with the -s argument followed by the rule name
For example trim the reads by running:
snakemake --printshellcmds \
--keep-going \
--use-conda \
--cluster "sbatch \
--time=60 \
--nodes=1 \
--partition=thin \
--cpus-per-task=32 \
--mem=64000 \
--mail-type=FAIL,TIME_LIMIT \
--mail-user=your@email.address \
--output=path/to/slurm_out/slurm-%j.out" \
--jobs 500 \
--max-jobs-per-second 3 \
--max-status-checks-per-second 5 \
-s trimming.smkRunning on a local machine is not recommended as some steps may exceed available
resources.
Rules of the pipeline are run separately. To run a rule, first change
directories into the rule's folder and run snakemake with the -s argument
followed by the rule name.
In the Demo we are running ITSFASTR on a local machine. Expected run time: ~ 1.5 h
First download the human reference genome (GRCh38_no_alt_analysis_set) and the mappability track. Extract the reference genome, and move the mappability track to /resources/Griffin_LR.
For the nucleosome positioning analysis we use nucleosome covered quiescent genomic sites (Quies)
and nuclosome free transcriptional start sites (TssA). Download files containg the genomic positions
of these sites from here.
Set the path to the files containing the sites of interest in the config file site_list.
Change directory: cd workflow/rules/1_preprocessing/
Remove sequencing adapters by running:
snakemake --printshellcmds \
--keep-going \
--use-conda \
--cores 8 \
-s trimming.smkAlign the reads to the previously set reference genome by running:
snakemake --printshellcmds \
--keep-going \
--use-conda \
--cores 8 \
-s mapping.smkChange directory: cd ../2_copy_number_analysis.
The ichorCNA tool requires the path to the ichorCNA script to be set in the config file. To set the path first run:
snakemake --printshellcmds \
--keep-going \
--use-conda \
--cores 8 \
-s ichorCNA.smk \
--conda-create-envs-onlyThis will create the environment and provide you with a path to it, similar to .snakemake/conda/103fe64931ac3076f3305793328dd331_. Paste this path into the config file after the conda_dir: parameter.
Now run:
snakemake --printshellcmds \
--keep-going \
--use-conda \
--cores 8 \
-s ichorCNA.smkChange directory: cd ../3_fragmentation.
snakemake --printshellcmds \
--keep-going \
--use-conda \
--cores 8 \
-s fragmentation.smkChange directory: cd ../4_FrEIA.
Extract the fragment ends by running:
snakemake --printshellcmds \
--keep-going \
--use-conda \
--cores 8 \
-s preprocessing.smksnakemake --printshellcmds \
--keep-going \
--use-conda \
--cores 8 \
-s FrEIA.smkChange directory: cd ../5_Griffin_LR.
snakemake --printshellcmds \
--keep-going \
--use-conda \
--cores 8 \
-s griffin_GC_and_mappability_correction.smksnakemake --printshellcmds \
--keep-going \
--use-conda \
--cores 8 \
-s griffin_nucleosome_profiling.smkPlease download the expected results from here. The expected results archive contains the expected folder structure, quality check results, the copy number profile, the fragment length profile, the final files fragment end sequence proportions, diversity and the nucleosome profiles of the dummy data set.
If starting from xenograft samples reads from the host and the graft need to be separated after trimming. To do this, first create the index files for both the host and the graft genomes. \
Then add the paths to the config file after RefPath_Primary and RefPath_Secondary. \
Finally change directory: cd workflow/rules/6_xenomapping/
and run:
snakemake --printshellcmds \
--keep-going \
--use-conda \
--cores 8 \
-s xenomapping_LR.smkThe deconvoluted and mapped reads are stored in the same folder as the output of the normal mapping is.
For usage of the ITSFASTR pipeline please cite: van der Pol et al. Real-time analysis of the cancer genome and fragmentome from plasma and urine short and long cell-free DNA using Nanopore sequencing (2022)