Phylogenetic tree for viruses

Question

[utsavnp]

Hi all,

I have full-length genomes of 19 isolates of the same virus. I would like to make a phylogenetic tree to understand the relatedness of these isolates. What is the best program for making a MSA and the best method (Neighbor-joining vs UPGMA) for making the tree?

Thank you for your help.

Utsav

[GenoMax]

You probably want something with a GUI. Try: http://megasoftware.net/

[apelin20]

If I may ask, what viruses? I wouldnt use NJ or UPGMA, I would use bayesian/ML analysis under a gtr model.

[Guangchuang Yu]

yes, using phyml or raxml with GTR model make more sense.

Answer 1

[konstantinkul]

Make MSA with mafft (http://mafft.cbrc.jp/alignment/server/) Find the best substitution model with MEGA Create phylogenetic tree using ML (maximum likelihood) methods with model defined in MEGA

Answer 2

[the_phylo_guy]

The VICTOR webservice (Virus Classification and Tree Building Online Resource) was precisely devised for that:

https://victor.dsmz.de

It is based on the following thoroughly prepared methods paper published in Bioinformatics:

https://doi.org/10.1093/bioinformatics/btx440

From the VICTOR web page:

This web service compares bacterial and archaeal viruses using their genome or proteome sequences. The results include phylogenomic trees inferred using the Genome-BLAST Distance Phylogeny method (GBDP), with branch support, as well as suggestions for the classification at the species, genus, subfamily and family level. The subfamily rank is only investigated during amino acid-based analyses as detailed in the VICTOR publication. (The service can be applied to other kinds of viruses, too, but has not yet been tested in this respect.) Upload your FASTA files, GenBank files and/or GenBank accession IDs below. The VICTOR methods and principles are described in the accompanying publication. For other details please see the VICTOR FAQ page.

Also please note: UPGMA is not an actual phylogenetic method and also assumes that your data has evolved under a molecular clock, which is rarely the case. UPGMA is considered as unreliable and this also applies to the bootstrap values of UPGMA. UPGMA might have a higher support for branches of the wrong tree than NJ for branches of the correct tree.

Regarding NJ, it is a phylogenetic method but it has been superseded by newer methods [1].

(VICTOR uses FastME for tree reconstruction.)

[1] https://www.ncbi.nlm.nih.gov/pubmed/18428768