This is a test version of Biostars. For the public version, visit https://www.biostars.org.
whole-genome sequencing of a non-model organism: from raw reads to genome assembly and gene annotation

I am looming into a workflow of whole-genome sequencing data generated using Oxford Nanopore Technologies (ONT) for a new/non-model organism, and I would like to understand the standard bioinformatics workflow from raw sequencing data through genome annotation.

Specifically:

  1. Raw data processing: What are the recommended steps for processing ONT raw reads (basecalling, quality control, read filtering, error correction, etc.) before genome assembly?

  2. Genome assembly: What is the recommended approach and software for assembling an ONT whole genome, particularly for a previously unsequenced organism?

  3. Chromosome assignment: Once the genome is assembled, how do we determine which assembled sequences/contigs correspond to which chromosomes? If there is no reference genome for the organism, how can chromosome-level scaffolding or chromosome identification be performed?

  4. Genome annotation: What is the standard workflow for structural and functional gene annotation in a newly sequenced organism? How are genes, transcripts, coding sequences, and other genomic features identified when there is no closely related reference genome?

  5. Validation: What approaches are recommended to assess assembly quality, completeness, chromosome-level continuity, and annotation quality?

For a non-model organism without an existing reference genome, are there any important additional steps or considerations that I should be aware of?

genomic sequencing
• 196 views • link
updated by shelkmike shelkmike • written by WUSCHEL WUSCHEL

1 answer

A detailed answer to your question would be very long, I'll give a rather simplified short answer:
1) Basecalling by Dorado with a sup ("super-accuracy") model. Trimming by Dorado.

2) Hifiasm is usually considered the best assembler for reads made with 10.4.1 Oxford Nanopore flow cells and basecalled with a sup model. Though I prefer Mabs-hifiasm (https://github.com/shelkmike/Mabs); it is my own parameter optimizer for Hifiasm that produces better assemblies for highly heterozygous genomes.
After you assemble the contigs, you need to polish them with shotgun short reads, for instance using POLCA.

3) The best method to scaffold contigs is to use Pore-C reads or Hi-C reads with YaHS.

4) A classical tool for structural annotation is BRAKER, but in my experience, ANNEVO is probably better. It's a recent neural network ab initio annotator.
For functional annotation, a simple tool that produces good results is PANNZER2. Upload a FASTA file with proteins to the PANNZER2 server, and in a couple of hours you'll have a functional annotation.

5) Methods of assembly quality control:
a) BUSCO analysis and calculate_AG sinaplots.
b) Calculation of N50.
c) Search for telomeric repeats. They should not be in the middle of chromosomes.
d) Visualization of a chromosomal contact map in Juicebox. If you see scaffolding errors, fix them.
The best method to control for the annotation quality is to calculate BUSCO completeness for proteins from the annotation and divide it by the BUSCO completeness calculated for the genome. This value should be very close to 1. Keep in mind that BRAKER4 by default artificially adds BUSCO genes to the annotation, which makes this method useless. If you annotate with BRAKER4, I recommend disabling this behavior.

Also, after you assemble the genome, you need to find short contigs that belong to contamination and to the mitochondrial genome.

[source for all of this: my own experience and reading lots of articles]

I recommend looking into Materials & Methods sections of 5 recent articles about assembly of some genomes published in journals with a high impact factor. Usually, this is a good way to learn about state-of-the-art methods.

If you have some questions, feel free to ask (though I'll be without internet access for 10 days starting from the day after tomorrow).

• 0 views • link
written by shelkmike shelkmike

Thank you very much.

• 0 views • link
written by WUSCHEL WUSCHEL

I must also mention a method that, in my opinion, is quite a popular mistake. Some authors fill gaps in scaffolds using tools like TGS-GapCloser. Gap filling is so error-prone (see Figure 2f in the article https://pmc.ncbi.nlm.nih.gov/articles/PMC10815858/) that I think it should be entirely avoided.

• 0 views • link
written by shelkmike shelkmike

Log in to answer this question.

More posts like this