OrthoMCL VS MEGA aligned clusters

Question

[3335098459]

Hi there,

I am new to bio-informatics and R. I have been reading the posts on this website from last few days and it has created some confusions to my previous knowledge.

1. What is the difference between the dendrogram generated by OrthoMCL and MEGA based clustal/muscle?

2. For strains related to same species, what could be the best option for generating their clusters and phylogram tree? (As far as I know, the evolutionary relationship between strains is not that much important)

3. Can we perform the clustering with OrthoMCL then take some random clusters and perform the alignment to get evolutionary relation?(This is just to save time; as aligning and clustering multiple genomes could take weeks.)

Kindly correct me. Also if you feel that my questions are childish please forgive me.

Thanks.

Answer 1

[Jean-Karim Heriche]

What you're trying to do and what kind of sequences you have is not clear so I'll give a generic answer.

The orthoMCL procedure builds a matrix of pairwise similarities between sequences then uses the MCL algorithm to cluster the sequences into groups of putative orthologs. MEGA builds a phylogenetic tree from a multiple sequence alignment. By definition, orthologs are genes whose last common ancestor is a speciation event and this can only be inferred from a phylogenetic tree. So if you care about finding orthologs the proper way involves building a phylogenetic tree. OrthoMCL is a shortcut that seems to work reasonably when dealing with eukaryotic genes.
Concerning point 3, you can indeed group sequences by similarity and then build a phylogenetic tree from the clustered sequences.

[3335098459]

I am trying to compare multiple genomes of different strains. As there is high number of protein sequences, I am looking for a less time-consuming protocol.

[lieven.sterck]

it all depends a little what your goal is. If you just need a preliminary view on the relationships between genes you can get away with interpreting the orthomcl dendrogram, if you need more in-depth results you will need to go further to phylogeny indeed.

That being said: you should not consider these analyses as separate things: orthomcl or any other protein clustering software is always the first step in a phylogenetic analysis. So first you generate protein clusters (= gene families) which you then align and build tree from. It makes no sense to randomly building alignments of genes (let alone build randomly trees), so firstly you need to delineate the gene sets you want to build tress form.

Answer 2

[Tm]

I have not used MEGA, but yes OrthoMCl can work best when your aim is to compare different genomes. It takes a complete set of protein of all orgainsms as input and performs blastp to compare each sequence with others. It generates separate group of similar sequences and thus helps you to check which particular protein is present all organisms or specific to a particular organism. Thus OrthoMCL provides the tree based on proteins they share.

Means, if your aim is comparative genome analysis and check for genes which are exclusively present in particular strain, responsible for a particular characteristic then you can use orthoMCL.

[3335098459]

Yes. you are right. My aim is to compare different genomes belonging to same species (different strains actually). Thank you for quick response and explanation.