27 September 2016 5 3K Report

I have called SNPs in 280 genomes and have identified sites where at least 1 individual possesses an alternate allele to the reference genome. I then concatenated all of these sites together for each genome, resulting in sequences of roughly 650,000bp. I'm hoping to create a phylogenetic tree using Phylip and a Neighbour-Joining method, but I'm having trouble picking the most appropriate model of evolution.

Partition Finder and JModelTest - both excellent programs for deducing best fit models - appear to both expect triplet codons, and do not offer an obvious way to adapt them to cope well with concatenated SNPs from across a genome. I wonder if anyone has any suggestions of how I might use either of these programs, or an alternative solution? Specifically, would using Partition Finder's 'Data Blocks' option allow me to get past this issue (e.g. Charset = Start - End/1')?

Many thanks,

Ian

More Ian Wilson's questions See All
Similar questions and discussions