cornetto

Cornetto

Cornetto is a method for iterative genome assembly using nanopore sequencing from Oxford Nanopore Technologies (ONT). This repository documents the Cornetto bioinformatics protocol and the Cornetto toolkit (a programme written in C and a collection of shell scripts). Cornetto toolkit also features some useful commands for evaluating assemblies generated from any other method.

Documentation: https://hasindu2008.github.io/cornetto

Publication and data

• Publication link
• Preprint link (bioxiv)
• Talk video link (London Calling 2025)
• Raw data and assemblies for the preprint

Please cite the following when using cornetto in your publications:

Gamaarachchi, H., Stevanovski, I., Hammond, J.M. et al. Targeted sequencing and iterative assembly of near-complete genomes. Nature Communications 16, 10406 (2025). https://doi.org/10.1038/s41467-025-65410-x

@article{gamaarachchi2025targeted,
  title={Targeted sequencing and iterative assembly of near-complete genomes},
  author={Gamaarachchi, Hasindu and Stevanovski, Igor and Hammond, Jillian M and M. Reis, Andre L and Rapadas, Melissa and Jayasooriya, Kavindu and Russell, Tonia and Yeow, Dennis and Hort, Yvonne and Patel, Chirag and others},
  journal={Nature Communications},
  volume={16},
  number={1},
  pages={10406},
  year={2025},
  publisher={Nature Publishing Group UK London}
}

Cornetto bioinformatics protocol

Detailed documentation

See here.

Summary

1. Generate a base assembly: Sequence on a single flowcell as normal and use hifiasm to assemble.
2. Generate a cornetto panel for adaptive sampling: Identify the ‘boringbits’ from the assembly and create a adaptive sampling target panel for rejecting reads.
3. Run adaptive sampling using readfish with the panel created in step 2 for around 24 hours.
4. Create the next iteration of assembly: Use collective data from step 1 and step 3 to create an assembly with hifiasm.
5. Repeat from step 2.

Cornetto toolkit

Detailed documentation

See here.

Summary

Using helper scripts:

scripts/telostats.sh asm.fasta # prints the telomere counts
scripts/minidotplot.sh ref.fasta asm.fasta # creates reference vs assembly dotplot. output: assembly.eps
scripts/asmstats.sh asm.fasta  # prints chromosome-wise assembly to reference report

Using individual commands:

# creating a dot plot
## 1. align assembly to reference
minimap2 --eqx -cx asm5 ref.fasta asm.fasta > asm.paf
## 2. fix the +/- directions to match the reference
cornetto fixasm asm.fasta asm.paf -r asm.report.tsv -w asm.fix.paf > asm.fix.fasta
## 3. dot plot
cornetto minidot asm.fix.paf -f 2 > asm.eps

# per-chromosome evaluation. Note: asm.windows.0.4.50kb.ends.bed is from `scripts/telostats.sh`
cornetto asmstats asm.paf asm.windows.0.4.50kb.ends.bed -r asm.report.tsv -s ref.fasta

# miscellaneous commands
cornetto fa2bed asm.fasta > asm.bed  # create a bed file with assembly contig lengths
cornetto seq -m 10000 reads.fastq > long.fastq # extract reads >=10kb

Acknowledgement

• cornetto uses klib (MIT licensed).
• cornetto minidot is simply minidot from miniasm (MIT licensed).
• cornetto sdust is simply sdust.
• cornetto telofind, telobreaks & telowin are implemented based on VGP telomere scripts (BSD licensed).

Fun fact

Regarding the name – the method is named by Ira Deveson after the ‘Cornetto’ ice-cream cone. There was a famous TV advertisement in Australia in 2000s that promoted the Cornetto as having ‘no boring bits’. In our cornetto workflow, we reject the ‘boring bits’, the bits which are too easy for the assemblers to resolve.

This site is open source. Improve this page.