QuickStart

Easy Install

noise2read is currently distributed on PyPI and bioconda, quick install noise2read in the build environment.

• Install via bioconda

conda install -c bioconda noise2read

conda install py-xgboost-gpu

Note

Currently, noise2read at bioconda requires installing py-xgboost-gpu separately. I will include this dependency in later release.

• Install via pip

pip install noise2read

and then install bioconda distributed packages of seqtk and bcool.

conda install -c bioconda seqtk bcool

Optional to install pygraphviz if you need the visualised read graph.

conda install -c conda-forge pygraphviz

Examples

1. General correction for short reads set with default parameters.

   • Training with CPU

   noise2read -m correction -i *.fa/fasta/fastq/fq -a True -d output_directory
   

   • Training with GPU

   noise2read -m correction -i *.fa/fasta/fastq/fq -a True -g gpu_hist -d output_directory
   

2. Correcting amplicon sequencing data with default parameters

   • Training with CPU

   noise2read -m amplicon_correction -i *.fa/fasta/fastq/fq -a True -d output_directory
   

   • Training with GPU

   noise2read -m amplicon_correction -i *.fa/fasta/fastq/fq -a True -g gpu_hist -d output_directory
   

Note

We strongly recommend utilizing GPU for model training and prediction, especially for large data sets, rather than using a CPU. If a GPU resource is available.