CONSENT

CONSENT (Scalable long read self-correction and assembly polishing with multiple sequence alignment) is a self-correction method for long reads. It works by, first, computing overlaps between the long reads, in order to define an alignment pile (i.e. a set of overlapping reads used for correction) for each read. Each read's alignment pile is then further divided into smaller windows, that are corrected idependently. First, a multiple alignment strategy is used in order to compute consensus. Then, this consensus is further polished with a local de Bruijn graph, in order to get rid of the remaining errors. Additionally to error correction, CONSENT can also perform assembly polishing.

Requirements

• A Linux based operating system.
• Python3.
• g++, minimum version 5.5.0.
• CMake, minimum version 2.8.2.
• Minimap2 available through you path.

Installation

Clone the CONSENT repository, along with its submodules with:

git clone --recursive https://github.com/morispi/CONSENT

Then run the install.sh script:

./install.sh

If you do not already have minimap2 available through your path, you can then run:

export PATH=$PWD/minimap2:$PATH

CONSENT should then be able to run.

Getting started

An example dataset (10x of simulated PacBio reads, raw assembly, and reference genome) is provided in the example folder.

Please run the following commands to try out CONSENT on this example.

Self-correction

To perform self-correction on the example dataset, run the following command:

./CONSENT-correct --in example/reads.fasta --out example/correctedReads.fasta --type PB

This should take about 2 min and use up to 750 MB of RAM, using 4 cores.

Polishing

To perform assembly polishing on the example dataset, run the following command:

./CONSENT-polish --contigs example/rawAssembly.fasta --reads example/reads.fasta --out example/polishedAssembly.fasta

This should take about 15 sec and use at most 150 MB of RAM, using 4 cores.

Running CONSENT

Self-correction

To run CONSENT for long reads self-correction, run the following command:

./CONSENT-correct --in longReads.fast[a|q] --out result.fasta --type readsTechnology

• longReads.fast[a|q]: fasta or fastq file of long reads to .
• result.fasta: fasta file where to output the corrected long reads.
• readsTechnology: Indicate whether the long reads are from PacBio (--type PB) or Oxford Nanopore (--type ONT)

Polishing

To run CONSENT for assembly polishing, run the followning command:

./CONSENT-polish --contigs contigs.fast[a|q] --reads longReads.fast[a|q] --out result.fasta

• contigs.fast[a|q]: fasta or fastq file of contigs to polish.
• longReads.fast[a|q]: fasta or fastq file of long reads to use for polishing.
• result.fasta: fasta file where to output the polished contigs.

Options

  --windowSize INT, -l INT:      Size of the windows to process. (default: 500)
  --minSupport INT, -s INT:      Minimum support to consider a window for correction. (default: 4)
  --maxSupport INT, -S INT:      Maximum number of overlaps to include in a pile. (default: 150)
  --maxMSA INT, -M:              Maximum number of sequences to include into the MSA. (default: 150)
  --merSize INT, -k INT:         k-mer size for chaining and polishing. (default: 9)
  --solid INT, -f INT:           Minimum number of occurrences to consider a k-mer as solid during polishing. (default: 4)
  --anchorSupport INT, -c INT:   Minimum number of sequences supporting (Ai) - (Ai+1) to keep the two anchors in the chaining. (default: 8)
  --minAnchors INT, -a INT:      Minimum number of anchors in a window to allow consensus computation. (default: 2)
  --windowOverlap INT, -o INT:   Overlap size between consecutive windows. (default: 50)
  --nproc INT, -j INT:           Number of processes to run in parallel (default: number of cores).
  --minimapIndex INT, -m INT:    Split minimap2 index every INT input bases (default: 500M).
  --tmpdir STRING, -t STRING:    Path where to store the temporary overlaps file (default: working directory, as Alignments_dateTimeStamp.paf).
  --help, -h:                    Print this help message.

Notes

CONSENT has been developed and tested on x86-64 GNU/Linux.
Support for any other platform has not been tested.

Authors

Pierre Morisse, Camille Marchet, Antoine Limasset, Arnaud Lefebvre and Thierry Lecroq.

Reference

Morisse, P., Marchet, C., Limasset, A. et al. Scalable long read self-correction and assembly polishing with multiple sequence alignment. Sci Rep 11, 761 (2021). https://doi.org/10.1038/s41598-020-80757-5

Contact

You can report problems and bugs to pierre[dot]morisse[at]inria[dot]fr