Is it possible to study or determine horizontal gene transfer or the signs of HGT in evolution using sequence alignment or any other bioinformatics/in silico tools?
If your goal is to identify inter-lineage hybridisation, then I suggest you use the topology-based parsimony approach implemented in PhyloNet. It uses (rooted) gene trees as input and aims at identifying hybridisation in the presence of incomplete lineage sorting, and quantifying the % of genes transferred. If your goal is to identify single genes that have been transferred horisontally, then this is more complicated as you need to use a method that takes into account coalescence times.
Sequence alignment is the key for describing gene transfer. You may want to look http://www.ncbi.nlm.nih.gov/CBBresearch/Koonin/ Koonin's work. The best is to define the problem and look for Gene Transfer. This may converge to some extend. FASTA all versus all was used by Walter Gilbert's Group for studying intron evolution. You can use similar methods. CD-HIT could be FAST at 20% similarity. It should be noted that the mechanism of gene transfer and hypothesis of gene transfer is not fully clear.
If your goal is to identify inter-lineage hybridisation, then I suggest you use the topology-based parsimony approach implemented in PhyloNet. It uses (rooted) gene trees as input and aims at identifying hybridisation in the presence of incomplete lineage sorting, and quantifying the % of genes transferred. If your goal is to identify single genes that have been transferred horisontally, then this is more complicated as you need to use a method that takes into account coalescence times.
Offcurse there are a lot of strategies.
The problem coming when you want to determine which is the donor organism and the fount organism. This apparently superfluous problem could have tremendous conceptual impact. e.g http://www.ncbi.nlm.nih.gov/pubmed/20551688
This paper is very intrensting because Forterre P. propouse a method to determine the direction of the HGT between viruses vs cells
@Thomas my approach is similar to the inter-lineage hybridization. By analyzing amino acid sequences, we have noticed distinctive differences between closely related homologs and intend to find the underlying reason for the same. HGT being the most commonly observed phenomena throughout evolution, we assume that it could be the reason for our findings. As you have suggested, would a maximum parsimony method offered by MEGA be useful?
@Manan: I guess no. The method in MEGA produces a tree while you are interested in a network. For that you'll need to use more sophisticated approaches. The problem is that there are few methods available and that they require the specification of somewhat non-intuitive parameters -- STEM requires that the effective population size is known, and PhyloNet requires that the number of hybridisations is known. I have been working on a way to detect and estimate past events of hybridisation based on coalescence distributions (paper is accepted for publication in Science), but this approach requires ultrametric trees (and considerable filtering of the genes used). I suggest you run PhyloNet for different numbers of hybridisations (0, 1, 2, 3, ...). At least that should give you an indication whether and where gene flow has occurred in your phylogeny, and where to go from there.
- Badges
- Science method
- Similar topics
- Biological Science
- Evolutionary Biology
- Phylogenetics
- Phylogenetic Analysis