Hi Daniel,

Let's assume you have a species tree for your bacteria, estimated using multiple genes or preferable an entire genome. Let's say you find out that species A, B, C and D cluster as so (in newick format):

((A,B),(C,D))

If two genes are paralogous, first of all you would clearly see some sequence homology allowing for a reasonable alignment to even take place. Second, you would find each paralog present in each species of bacterium, unless one or both genes have been lost over time. 

Then, if you made an alignment of all of your genes of interest (i.e., putative paralogs), you would get a tree that generally mimics the species tree but with each paralog clustering together. With paralogs 1 and 2, you would have a tree as so:

(((A1,B1),(C1,D1)),((A2,B2),(C2,D2)))

This tree suggests that a gene duplicated in the ancestor of A, B, C and D, and that speciated into four species, each of which retained both paralogs.

If a gene was deleted from the genome or there is simply missing sequence data (i.e., the database doesn't contain it for a species), you might get a tree that resembles this:

(((A1,(C1,D1)),((A2,B2),D2)))

Hope that helps!

-Chris

First of all you must to decide what aim of your phylogeny and what level of it you interested in. As I can understand from your question, you have not enough varible sequences for obtaining clear phylogenetic structure? What algorythms and techniques you used for this?

Note that paralogous genes are homologous genes that occur within one species and have diverged after a duplication event. Unlike orthologous genes, a paralogous gene is a new gene that holds a new function. These genes arise during gene duplication where one copy of the gene receives a mutation that gives rise to a new gene with a new function, though the function is often related to the role of the ancestral gene. So, if you know the function you may potentially estimate presence/absence of these genes.

I think it woud be at least two solutions of your problem:

-building the tree as supertree on this both genes or with adding smth. new ones

-searching for more varible markers

Hi, Andrii. 

The two genes I am working with are RecA and the putative paralogous, RadA/Sms (it's not same from archeae).

When I align RadA using the refseq database from Ncbi I get many sequences that are closely related. I think to infer a paralogy relationship between two genes I'd need to make a tree using more distant sequences, don't you agree?

I understand the concept of paralogous genes, I just don't know how to prove (or not) if one gene (RadA/Sms) appear as a result of an event of duplication.

I was thinking in build a tree with as many sequence of RadA as possible together with several sequences of RecA. Then, I'd get a branch of RadA sequences that are very close to a branch of RecA and make a new tree with these two groups.

What do you think? Do you suggest another strategy?

Thank you for helping. 

Hi Daniel,

Let's assume you have a species tree for your bacteria, estimated using multiple genes or preferable an entire genome. Let's say you find out that species A, B, C and D cluster as so (in newick format):

((A,B),(C,D))

If two genes are paralogous, first of all you would clearly see some sequence homology allowing for a reasonable alignment to even take place. Second, you would find each paralog present in each species of bacterium, unless one or both genes have been lost over time. 

Then, if you made an alignment of all of your genes of interest (i.e., putative paralogs), you would get a tree that generally mimics the species tree but with each paralog clustering together. With paralogs 1 and 2, you would have a tree as so:

(((A1,B1),(C1,D1)),((A2,B2),(C2,D2)))

This tree suggests that a gene duplicated in the ancestor of A, B, C and D, and that speciated into four species, each of which retained both paralogs.

If a gene was deleted from the genome or there is simply missing sequence data (i.e., the database doesn't contain it for a species), you might get a tree that resembles this:

(((A1,(C1,D1)),((A2,B2),D2)))

Hope that helps!

-Chris

Thank you, Christopher.

Hello Daniel

Usually phylogenetic analysis are used to detect Horizontal Gene Transfer (HGT), where species tree (eg- 16s rRNA tree is used fro bacteria) with gene tree

T-Res is tool for such analysis (one of the best)

http://www.trex.uqam.ca/index.php?action=hgt

But, if you are interested in paralogous detection then it is recombination you are looking for

In this case

Try "Single Break Point Recombination" Test using Hyphy

http://hyphy.org/w/index.php/User_registration_and_downloads

Or

Use RDP (you will find and compare many algorithms here)

http://web.cbio.uct.ac.za/~darren/rdp.html

Here is one of my paper where i have used such tools and methods:

Article Identification of the positively selected genes governing ho...