I need to construct the phylogenetic tree of the whole genus of streptomyces. And my goal is to find out how my target strep species distributed in the whole strep genus and obtain the distribution pattern. Strep is a big genus, has 658 species(65% of them have the genome info), and my target species is about 25. But the traditional 16S method seems not helping when it comes to strep is a genetic recombination genus. And still lack of MLST scheme to distinguish my target species. I want to get the phylogeny for the whole genus, but I lack related genetics knowledge.

There is something I did before try to get the tree.

  1. I ran the orthomcl program for 400 strep genomes which I got from NCBI, and the result is: I got 22 orthologues
  2. I ran the fastani program for 200 strep genomes(type material) and got the ani result.

What can I do to achieve my goal? Thanks!

  • $\begingroup$ Hi @Tangli83, I will construct an answer in tomorrow evening (Saturday morning in China) based on our conversation $\endgroup$
    – M__
    Mar 12 '20 at 14:12
  • 1
    $\begingroup$ Oh, thanks,~ I thought my post didn't explain clearly. $\endgroup$
    – TANGLi83
    Mar 12 '20 at 16:03
  • $\begingroup$ that's fine I know your questions and research aims $\endgroup$
    – M__
    Mar 13 '20 at 4:37

You need bootstrap support using a model-based tree building algorhithm, via maximum likelihood (a few people use Bayes). The file format is relaxed phylip format (please submit a separate question if you have difficulties here - its a bit tricky).

I use standard RAxML here, specifically raxmlHPC (easily downloadable and compiles on Linux and OSX). The codes are quite complicated and I've given them below.

A robust maximum likelihood tree is,

raxmlHPC -m GTRGAMMA -n output.tre -o outgroup-p 10000 -s alignment_file.phy

To make a bootstrap tree,

raxmlHPC -b 459 -m GTRGAMMA -n output.tre -o outgroup -p 10000 -s /path_to_alignment.phy -N 500

This tree will bootstrap for 500 replications, however to start I would use 100 replications.

Make a consensus tree of the bootstraps,

raxmlHPC -f b -m GTRGAMMA -n output.tre -o outgroup -p 10000 -t /path_to_master_tree(best_tree) -z path_to_bootfile

You require bootstrap support >80% and please repeat this with and without 5-2/5-3 (it still looks long)

The cluster you have access to is fine for the calculation, it will take aound 24 hours to complete one bootstrap calculation for one data set and obviously you need to parallelise you calculation across the 22 contigs.

Viewing the tree, FigTree (for Mac OSX) is easy.


Rooting can be complicated because I don't really know your bacteria.

The recombination issue is more complicated, but I would construct 22 trees from your contigs and assess them for congruence. Panmixia is a concerned, which means too much recombination


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.