I retrieved 1500 sequences (from NCBI) of 6 different protein isoforms belonging to metazoans. Now, the issue is if it is possible to deal with such a large number of sequences in order to reconstruct the phylogenetic tree. What could be a clever strategy? Does this number pose challenge for dedicated software like PhyML, MEGA or MrBayes? What kind of these software should I use?
I can suggest RAxML (http://sco.h-its.org/exelixis/web/software/raxml/index.html). I think it can handle such numbe of sequences.
Certainly lot of software will handle 1500 sequences. The question is whether the results will be interpretable or reliable. I would think that across such a large group, that the evolutionary processes will not be similar. In other words finding a model of sequence evolution that fits such a large variety of taxa could be difficult. And yet if the proteins are reasonably conserved it may work. I would start out with a neighbor joining tree to see if the relationships make any sense and to check that the alignment for so many taxa is optimal. Then, for the search, use maximum likelihood methods in Phyml, which is fast and reliable. The implementation in Seaview is handy since you can align and optimize in the same program.
You can try our MetaPIGA software, a robust implementation of several stochastic heuristics for large phylogeny inference (under maximum likelihood). You can download the software here: http://www.metapiga.org
Just an advise: make sure that using 1500 sequences makes sense. For example, do you have enough informative sites to resolve all internal nodes? In other words, a high ratio of taxa number versus number of variable sites is not good.
You can use NJ (very fast) to identify near-zero branch lengths and use this analysis for guiding your choice of taxa for ML or Bayesian analyses.
I did not know metaPIGA; this is a good advice. As well as the use of NJ as initial trials
Well, try MetaPIGA, I'm sure you will like it. If you have any technical question, feel free to contact Djordje (the PhD student developing the software) at Đorđe Grbić
Best, M.
I'm surprised that I did not suggest MetaPiga in the first place. Terriffic program that I use all the time. The model section feature is one of the best in the field and the versatility - and fun - of the analyses are handy.
Thank you Patrice for your interest in MetaPIGA. We are continuing its development for yet increased speed and versatility. Keep an eye on www.metapiga.org, version 4 will come out soon (we have also implemented MC3 in version 4).
Best Regards.
-------------------------------------------------------------------------------------------------
Prof. Michel C. Milinkovitch
Laboratory of Artificial & Natural Evolution (LANE)
Dept of Genetics & Evolution | University of Geneva
Science III, quai Ernest Ansermet 30
1211 Geneva, Switzerland
Tel +41(0)22 379 6785; Fax +41(0)22 379 6795
Visit www.lanevol.org
News at: www.lanevol.org/LANE/News/News.html
Visit www.metapiga.org and try MetaPIGA v3, a software for large phylogeny inference
EpiPhysX: The Physics of Epithelia - check the movie here https://goo.gl/tXiGmx
How chameleons change colour - check the movies here https://goo.gl/tEUG5L
R2OBBIE-3D: scanning Life with high resolution - check the movies here https://goo.gl/OmDKF5
-------------------------------------------------------------------------------------------------
My phylogenetics students seem to understand the concept of a tree search easily with this program. Looking forward to the next version.
Another option is that you can use the web site of CIPRES, The CIPRES Science Gateway is a public resource for inference for large phylogenetic trees. It is designed to provide all researchers with access to large computational resources of the NSF TeraGrid through a simple browser interface. You can deal with your proteins and run your analysis in several closters at the same time, also you have the option to make many kind of differenta analysis: RaxML, Bayes, Create Multiple Alignments from Sequences, NJ trees, JModeltest, using different partitons, dating usin BEAST, etc, etc. I reccomend this web site: http://www.phylo.org/sub_sections/portal/
The best part is that you dont need to use RAM memory of your computer because all the analyses as I told you are doing in the cluster in the web.
Good luck! :)
- Badges
- Science topic
- Similar topics
- Biological Science
- Immunology
- Antibodies