Hi Divid,

    If the phylogenetic relationship is your major concern, I would suggest you infer the phylogenetic tree using maximum likelihood and/or Bayesian methods. For maximum likelihood method you can use RAxML and for bayesian you can use MrBayes among many programs. See http://evolution.gs.washington.edu/phylip/software.html for a comprehensive list. I wold not suggest you to trust the guide free because it is used to facilitate alignment and in most cases is NOT reliable from a phylogenetic perspective. 

  See review paper Yang Z & Rannala B 2012. Molecular phylogenetics: principles and practice. Nature Review Genetics 13: 303-314, to get started. 

http://www.nature.com/nrg/journal/v13/n5/full/nrg3186.html

 http://abacus.gene.ucl.ac.uk/ziheng/pdf/2012YangRannala.NRGv13p303.pdf

Hope this helps.

Thank Chengmin. Now my problem is that I don´t know how to convert a XMFA output file to RaxML input. Do you know any way to do this?

I would question why you would want one "best tree" of 30 strains or isolates within one species.  I suppose it depends a bit on how those strains or isolates were sampled, but in general there has been a lot of horizontal gene transfer, via phages and plasmids and other mechanisms, moving genes around not only within the Escherichia but also between Escherichia, Shigella, Salmonella and other enterobacteria.  This does not mean that phylogenetic analyses are not very useful, but it is often more useful to analyze individual genes or groups of genes separately.  Whether doing the genes separately or constructing a whole genome phylogeny, it can be difficult to compare the right copies of multi-copy genes,  If there are 2 or more copies of a given gene within the genome of each isolate, it may not be obvious which copy in isolate A to compare to which copy in species B.  It could be very useful to construct one "best tree" using a set of core genes such as DNA polymerases, ribosomal proteins and rRNA genes to which trees built from other genes could be compared individually or in groups.

There are quite a few software packages which build "webs" or networks which give a sort of representation of gene flow between groups or isolates, but for the most part they do now show which genes moved exactly, or the direction of gene flow.  Nor can they answer questions such as "do genes closer to the origin of replication tend to be less mobile than genes further from the origin?".

The XMFA format is a modification or extension of FASTA format. http://darlinglab.org/mauve/user-guide/files.html  Most phylogenetic analysis programs can read FASTA format, and there are many conversion programs for converting between common formats, but I am not aware of one that can deal with XMFA because it would need to know how to deal with multi-copy genes and other issues.

Hi Brian,

Yes, you right. Horizontal mobile elements are a concerns when the whole genome is compared in species like E. coli. Many investigators prefer to do MLST in silico to avoid possible interferences caused by horizontal gene transfer. Indeed, right now I am using only the "'soft core" (about 3000 gene) [see: http://www.biomedcentral.com/1471-2164/13/577] of E. coli for do the comparison and not the whole genome. Thanks for your answer.

An approach you might want to consider trying, is one similar to the methods used in this paper:

Rapid phylogenetic and functional classification of short genomic fragments with signature peptides

http://www.biomedcentral.com/1756-0500/5/460

For a set of bacteria within the same Genus, you would probably need DNA rather than amino acid k-mers and a much longer k-mer length (maybe as much as 1,000 bases, as opposed to 7 to 9 amino acids), in order to cluster your data into clades.  I am guessing that this could be a quick way to see which gene regions are found only in Escherichia coli, which are shared between Escherichia, Salmonella, Shigella, and which seem to have come from outside the Escherichia/Salmonella/Shigella group.  I think you would input the genomes of many enterobacteria along with your Escherichia genomes, for this type of study.