Sidenote: Eduardo doesn't have to worry about degeneracy if he has both nucleotide and protein sequences. He knows which aa matches which codon.

Hi Eduardo,

Try using PASTA to align and RAxML for tree estimation. You might want to consider doing individual gene trees and incorporate some other data from GenBank so you can test hypotheses on horizontal gene transfer.

Best of luck!

Adam

http://www.ncbi.nlm.nih.gov/pubmed/25549288

http://sco.h-its.org/exelixis/web/software/raxml/index.html

http://www.cs.utexas.edu/~phylo/software/pasta/

Hi,

In my opinion your analysis should be based solely on a single molecular phylogenetic approach...instead in addition to NJ or UPGMA analysis also do ML and MP analysis. Back translation may pose some problems as there are more than one codon for many amino acids..

Sidenote: Eduardo doesn't have to worry about degeneracy if he has both nucleotide and protein sequences. He knows which aa matches which codon.

Hi Eduardo,

Try using PASTA to align and RAxML for tree estimation. You might want to consider doing individual gene trees and incorporate some other data from GenBank so you can test hypotheses on horizontal gene transfer.

Best of luck!

Adam

http://www.ncbi.nlm.nih.gov/pubmed/25549288

http://sco.h-its.org/exelixis/web/software/raxml/index.html

http://www.cs.utexas.edu/~phylo/software/pasta/

Adam, Can it be also used for bacterial genome alignment!

I've never personally used it for bacterial genome alignments -- only individual genes. But, I have colleagues who have used it for whole chloroplast alignments, so I think it will work for bacterial genomes. 

If whole genome alignment gives you problems try aligning each individual gene, then concatenate them to make a "supergene", and build a tree from this supergene. I am not too sure what your goals are for a whole genome alignment, but if you are after some sort of hypothesis testing of relatedness between species then the supergene method should work. A slight variation to this method would be to create individual gene alignments and use the program BEAST to estimate a species tree -- BEAST takes in individual genes to estimate a multilocus tree or species tree.

http://beast2.org/

Adam, Thank you for all you feedback, am investigating the resources you mentioned after this msg. So far what I have done is to create separate alignments using both the back-translated protein sequences and the concatenated protein sequences just to compare. What troubles me is that although the alignments look good for either method, the trees vary substantially. Am treating the data in the same way and the clustering varies. Am not sure why. 

If the bacterial species in your study are distantly related, it might be a better idea to use aa sequences rather than DNA sequences.

It is not clear why you are using UPGMA.  It certainly is fast, but unless the assumption of a molecular clock is met, it could return an incorrect tree.  Computers and the algorithms used today are very fast.  Why not use ML?  First determine the model of nucleotide (or amino acid) substitution and then infer the tree using ML.  That would be my suggestion.

UPGMA and NJ both are distance method, UPGM can be used when the rate of gene substitution is more o less constant. NJ have nice statistical properties and result in a single tree. I suggest using more than one approach such as NJ, ML and MrBayes, the result from the same data with different approach should help you to know the homology and orthology of your sequences if you have enough data to find the most recent common ancestor.

Sudhindra R Gadagkar and Obdulia Segura-Leon  thank you very much for the feedback. I finally got a decent-looking tree using back-translated protein sequences and I set much higher open gap cost vs to gap extension (50:1) and treating the gaps like any other for the alignment and used NJ + 100x bootstrap method for tree. I've attached an image of the tree, and the sequences distantly clustered almost perfectly by species (S. pyogenes and S. Dysgalactiae) except in one instance (i.e., FCT-6). Could this be evidence or  support for horizontal gene transfer? Please let me know if you cannot see imagine, because I would really like to know that am interpreting the data correctly. I will however adhere to advice and try ML approach next for comparison. Again thank you for all the support thus far.