Can I use snp data for iqtree?

Question

[ylkim9]

Hello, I have some confusions using iqtree.

I have 20 fish individuals variant called vcf file using reference genome only in contig level.

I filtered it and changed it into a fasta file.

I tried alignment to make input file for iqtree but it failed cause this file too big(about 28GB) and I didn't have that much memory in root folder of linux server.

Do I have to extract some SNPs(e.g neutral SNPs...) and try it again?

Also I have confusions of process for making phylogenetic tree with it.

Variant calling -> Variant filtering -> convert vcf file to fasta file -> Annotation with reference genome -> MSA(I will use MAFFT) -> iqtree

Is it right? I have had too much errors to forget the basic process. Someone help me.

Answer 1

[Michael]

You don't need to align the complete genomes if you want to base the tree on variant calls and if you restrict the variants to SNPs (you should do that if you haven't already) as they are already aligned by genome position. When you convert the VCF to FASTA, ensure you have control over the allele inserted. It could be either the alternative or reference allele, the outcome will depend on it. If the file is still too big, extract only the variant positions. When you feed the input to IQtree, ensure it chooses the right model. IF the input consists only of invariant positions, tell IQTree that by specifying a model containing '+ASC'. Otherwise, you can let IQtree select the model or try GTR (or GTR+ASC) for starters.

If you further want to restrict the analysis to neutral sites it gets more difficult. It's been controversial which sites are "truly neutral". You could start by restricting the analysis to 4-Fold degenerate sites in coding regions. Such can be annotated with tools like degenotate if you have a good genome annotation.