What to do after BUSCO to build a phylogeny?

Question

[Nattapol]

Hi all, Forgive me if the question is silly. I'm a Ph.D. student and a little stuck in my bioinformatic pipeline. I am working on a phylogenomic paper on fungi. I assembled my genome and use BUSCO to pull out single-copy complete orthologs. Now I have ~ 1500 orthologous genes per taxon (each gene on a separate fasta file). There are 50 taxa in the dataset. I'm not sure what to go from there. I'd like to use ASTRAL but do not know how to make gene trees when you're analyzing each taxon separately. Like for example,

Taxon A has gene1, gene2, gene3 etc. in one folder

Taxon B has gene4, gene5, gene6 etc. in another folder

Taxon C has gene7, gene8, gene9 etc. in yet another folder and so on....

How do I know if gene1 and gene7 are the same? Do I have to concatenate the genes of each taxon first and blast all-v-all? I use the same marker set (odb10) for all taxa. Is there a program that will compare genes from all taxa and give me the consensus genes that all taxa share? Because I'm sure many genes are missing from certain taxa, or I shouldn't worry about that?

Any advice would be helpful! Thanks.

Answer 1

[patrickdm]

Hello, if working with protein sequences is an option, you could use OrthoFinder in order to extract the single-copy orthologue sequences shared by all your OTUs.

Please note that you will need the fasta files of the aminoacidic sequences for your genome's predicted genes and all the other taxa in the dataset; you can use Augustus to perform gene prediction on the assembly.

When you obtain the orthologues sequences clusters for your dataset, you'd want to align each cluster indipendently, then concatenate the sequences for each OTU in a single multifasta supermatrix and use it -i.e.- to infer a ML tree using IQtree2, or any other tree reconstruction algortihm and software of your choice.

Hth.