This repository is now unsupported and we do not recommend its use. Please contact Oxford Nanopore: [email protected] for help with your application.

ONT assembly and Illumina polishing pipeline

This pipeline performs the following steps:

• Assembly of nanopore reads using Canu.
• Polish canu contigs using racon (optional).
• Map a paired-end Illumina dataset onto the contigs obtained in the previous steps using BWA mem.
• Perform correction of contigs using pilon and the Illumina dataset.

If you only have nanopore reads consider using the katuali pipeline for assembly and polishing.

Getting Started

Dependencies

• Canu
• samtools
• BWA
• cmake
• racon - the pipeline will download and build it
• minimap2 and miniasm
• pilon - the pipeline will download it

Usage

Edit config.mk to set input files and parameters. Specifying the following is mandatory:

• NANOPORE_READS - input nanopore reads (note that this must be a single valid fastq file, see here how to combine fastq files).
• ILLUMINA_READS_PAIR1 - fastq with the first reads of the paired-end Illumina dataset.
• ILLUMINA_READS_PAIR2 - fastq with the second reads of the paired-end Illumina dataset.
• CANU_GENOME_SIZE - genome size parameter passed to canu.
• PILON_MAX_MEM - maximum amount of memory used by pilon. Increase this is if pilon crashes because of running out of memory.

The number of cores used can be specified by CORES (set this to the number of CPUs in your machine). Racon corrections can be disabled by setting USE_RACON=no. Pilon polishing can be disabled by setting USE_PILON=no.

Then issue issue make all to run the pipeline. Issue make help for a list of utility make targets. Issue make clear_wdir to delete the working directory (including all results!).

Using through docker

The easiest way to use the pipeline is through docker. First install docker, then issue the following to build the container:

cd docker; make build

Then run the container:

docker run -v /path/to/my_data:/data -it ont-assembly-polish

You will be dropped into the directory /home/ont-assembly-polish, then simply edit config.mk and run the pipeline. The -v flag will make the /path/to/my_data directory on the host available under /data in the container.

Results

Evaluation on simulated data

In order to evaluate the performance of the pipeline we have simulated long and short reads from the yeast genome and measured the accuracy of recovered contigs after various stages of correction.

Long reads were simulated using an in-house script under the following conditions:

• Number of reads: 150000
• Read lengths sampled from a gamma distribution with mean 6000 and shape 0.5 and a minimum read length of 600
• Simulated error rate was 0.1, errors were uncorrelated events of size one with substitution:insertion:deleltion ratio of 1:1:2

Short reads were simulated using simLibrary and simNGS:

• Simulated Illumina data consisted of paired-end reads of size 101, with the default insert length of 400
• Simulation runfile can be found under: data/s_1_4x.runfile
• The number of simulated read pairs was 21666129 (360x fragment coverage)

We have measured the accuracy of recovered contigs after various correction stages using dnadiff from the mummer package and last:

Conclusions from the evaluation:

• At the simulated error rate, canu alone recovers high accuracy contigs.
• Both dnadiff and lastal accuracies suggest that racon and pilon polishing increases contig accuracy.
• Lastal accuracies suggest that the effect of racon and pilon polishing is additive, though the increase in accuracy is not substantial.

Help

Licence and Copyright

(c) 2016 Oxford Nanopore Technologies Ltd.

This Source Code Form is subject to the terms of the Mozilla Public License, v. 2.0. If a copy of the MPL was not distributed with this file, You can obtain one at http://mozilla.org/MPL/2.0/.