de novo chromosome level assembly

Question

[mthm]

I have the contigs assembly of a non model species. the assembly itself is made de novo since there is no close reference genome for this genus. I have used RagTag before for chromosome level assembly but it requires reference chromosome assembly, which in this case, is from a very distant genus. What would be your suggestion, do I use that distant reference chromosome or run de novo chromosome assembly? and in the latter case, I found CSA (Chromosome-Scale Assembler), are there better suggestions in terms of robustness?

[colindaven]

Plant, animal, expected genome size ?

[mthm]

Thanks for reply, it is an insect genome, expected genome size 1-2 Gb

[lieven.sterck]

what kind of data types do you have at hand?

Answer 1

[lieven.sterck]

I would personally NOT use the distant one to do reference guided scaffolding, the risk that things have changed (chromosome wise) over such an evolutionary distance is just too high to rely on the distant reference.

My suggestion would thus be to go for denovo scaffolding (if feasible off course, == if you have the data for it). Depending on your available data there is likely several options so best to first fill us in on what data you might have before we can do tool suggestion.

[mthm]

Thank you for your answer. Data are whole genome sequencing using both short reads (Illumina hiseq 10) and long reads (PacBio) techniques. The genome is done using a hybrid assembly with wtdbg2 and MaSuRCA hybrid assembler following polishing the data and removing contaminants. My data is from a cricket family and you are right, the genome size, the chromosome numbers and even the sex chromosome mechanism varies a lot in this suborder.

[colindaven]

I would reassemble with hifiasm and polish if needed with the short reads. wtdb2 is known for creating very short fragmented contigs.

[mthm]

Thanks for the suggestion, I first thought it is for contig assembly but apparently can also be used for scaffolding at chromosome level

[mthm]

any tool suggestions would be appreciated

[lieven.sterck]

colindaven already suggested HiFiasm, perhaps you can also have a look at flye (it used to have a mode to even do ref-guided assembly, but not sure if that's still supported?)

[mthm]

How can I attribute the scaffolds to chromosome numbers?

[lieven.sterck]

you mean assemble them into pseudo-chromosomes or identify which scaffold is derived from which 'real' chromosome?

[mthm]

I suppose the first one, since I don't have the information of the real chromosome

[colindaven]

Look at Ragtag on github, assuming you have a close reference to align against. Else your main option is spending more on Hi-C data. Else maybe optical mapping with Bionano.