How do I use phylogenetic analysis of BLAST search results to choose the best template for homology modelling?

Question

[roonymoonlight]

Hi,

I am trying to generate the best possible structure for an orphan GPCR by homology modelling. I have the sequence and BLAST results. I want to choose a template based on phylogenetic analysis. What phylogenetic software should I use? Does anyone know any resources I can use to guide me?

Thank you Roween.

Answer 1

[jhc]

A relatively easy way would consist of: 1) fetch the sequences from all the good matches in your blast results (i.e. passing standard evalue and coverage thresholds) 2) try any of the predefined workflows in ete-build.

Those workflows take care of all necessary steps with just one command line (from building multiple sequence alignments to phylogenetic inference. Even a preliminary tree image is returned). You can fine tune the steps or switch what software is used for each task, but something as simple as this would allow you to run the same pipeline used in eggNOG and PhylomeDB phylogenetic databases (which include testing multiple aligners, trimming, model selection and maximum likelihood inference):

ete3 build -w eggnog41 -a sequences.faa -o output_tree

If interested, things can get much more complex :)

[roonymoonlight]

Thank you very much,

How about these programs below: Phylip http://evolution.genetics.washington.edu/phylip PAUP* http://paup.csit.fsu.edu/about. MrBayes the package for Bayesian trees MEGA http://www.megasoftware.net/ I am a beginner and I need an advice, which program is the best ? and how these program are different from ete build ?

[jhc]

there is no best, they just use different approaches. I would say Bayesian and Maximum Likelihood are among the preferred methods nowadays. Note also that ete-build is not phylogenetic inference software by it self. It just guides the execution of different programs. The example workflows I mentioned above use several aligners (clustalOmega, mafft, muscle), trimmers (trimAl), modeltesters (pmodeltest) and tree inference programs, which are based on Maximum Likelihood calculation (Raxml, Phyml, Fasttree).