ONT Basecalling / Demux Pipeline

Small pipeline to perform basecalling and demultiplexing (optional) of ONT data, collect QC metrics and generate a MultiQC report. It uses Dorado for basecalling and demultiplexing.

Requirements

• Nextflow (>= 22.04)
• Apptainer / Singularity
• Dorado (0.5.3 tested). It can be used via container, or installed locally from https://github.com/nanoporetech/dorado.

Usage

• Clone this repository
• If you want to demultiplex: create a samples.csv file with at least the barcode and sample columns. The barcode column should contain the barcode used for demultiplexing (with the leading zero, e.g. barcode01), and the sample column should contain the sample name (this name with be used on the report and as name for FASTQ file).
• Copy params.example.yml (for example to ./my_params.yml) and modify it according to your needs. Remember to point sample_data parameter to the file created at the previous step.
• Run the pipeline passing your params file to -params-file option:

nextflow run ont-basecalling-demultiplexing/ -profile apptainer -params-file my_params.yml

Parameters

Parameter	Required	Default	Description
experiment_name	False	-	Name of the experiment, used for final reports (title and filename).
data_dir	True	-	Path to the folder containing the POD5 files.
sample_data	True	input/samples.csv	Path to the CSV file containing the sample data (required if demultiplexing).
output_dir	False	demultiplex_results	Path to the folder where the results will be saved.
fastq_output	False	true	If true, the pipeline will generate FASTQ files (if not, it would be UBAM files).
qscore_filter	False	10	Minimum QScore for the "pass" data, used for demultiplexing.
dorado_basecalling_model	False	dna_r10.4.1_e8.2_400bps_sup@v5.0.0	Model used for basecalling.
dorado_basecalling_extra_config	False	-	Extra configuration for Dorado basecalling.
dorado_basecalling_gpus	False	1	Number of GPUs to use for basecalling.
skip_demultiplexingskip_demultiplexing	False	false	If true, the pipeline will not perform demultiplexing
dorado_demux_kit	False	EXP-NBD196	Kit used for demultiplexing.
dorado_demux_both_ends	False	false	If true, the pipeline will demultiplex using barcodes from both sides (5' and 3').
dorado_demux_extra_config	False	-	Extra configuration for Dorado demultiplexing.
dorado_demux_cpus	False	16	Number of CPUs to use for demultiplexing.
use_dorado_container	False	true	If true, the pipeline will use Dorado via container (~3.5GB download). If false, it will expect to find it locally.

Considerations

• It is possible to run the pipeline either in SLURM clusters using -profile slurm.
• Basecalling and demultiplexing are performed on separated steps to allow for a better control of the resources used by each process, and to prevent a whole basecalling redo in case of a failure during demultiplexing, wrong kit specified, etc.
• The basecalling process uses GPU, so make sure to have one available. If using SLURM, the job will be submitted with --gres=gpu:X option.
• Demultiplexing step won't use GPU, only CPU.