Dear Azeem:

I would suggest the program RDP3 (Martin D, Rybicki E (2000) RDP: detection of recombination amongst aligned sequences. Bioinformatics 16: 562–563.) which Dr. Martin modified for our uses to be able to do whole bacterial genome alignments. We published this modified method together (Hiller, N.L., Ahmed, A., Powell, E., Eutsey, R.E., Earl, J., Martin, D., Janto, B., Hogg, J.S., Boissy, R., Barbadora, K., Post, J.C., *Hu, F.Z., and *Ehrlich, G.D. Generation of Genic Diversity among Streptococcus pneumoniae Strains via Horizontal Gene Transfer during a Chronic Polyclonal Pediatric Infection. PLoS Pathogens 6(9): e1001108, 2010).. Since it works for bacterial genomes I don't think you'll have trouble with viruses

Dear Claudio

I actually have used MAFFT for alignment when CLUSTAL and MUSCLE options were not working. Now I have this alignment file with me but still I am unable to search nucleotide substitution model via MEGA and jmodeltest and tree construction. Could you please tell me if there are any online resources for these steps?

Dear Azeem,

I may be wrong, but I think you are going to have a hard time finding a free online server to undertake such gargantuan task; I wonder if you can't have access to a server from some neighbour lab. Moreover, MEGA might not be the best option for such huge dataset. Have you tried the last HPC jModeltest version with PhyML for Linux servers?

If you have troubles with ModelTest, just run RAxMl on a proper server with GTR+Gamma.

Cheers

Jordi

Try Clustal Omega designed for large data sets.

http://www.ebi.ac.uk/Tools/msa/clustalo/

A dataset as large as yours cannot be analyzed using freeware. We have moved to the Cloud environment to conduct these analyses. It is relatively inexpensive if you don't use it for data storage.

Good luck

I agree with Jordi, MEGA won't be able to handle such a big dataset. RAxML will be able to provide a ML tree and you can try PhyML (can also work in parallel).

If you need to test models, you can use MrAIC or pMrAIC, which are based on PHYML and can run in parallel. (http://www.abc.se/~nylander/) On both you will need to tackle linux servers, ask for help to your sysadmin if you can.

If you can use an approximate approach you may also try FastTree using a GTR model.

Hi,

I think you need something designed for large datasets. I personally use Mauve Genome Alignment as it is very dependable from virus to bacterial genome datasets. Good Luck

Additionally, If you're having trouble running your data set on your computer, you could consider using Cipres Science Gateway clusters (http://www.phylo.org/portal2). They don't have all software you could wish for, but they have MAFFT, MUSCLE (for sequence alignment) and MrBayes and RAxML (for Bayesian an ML tree estimation). I have been working on bigger data sets using these software programs and the remote clusters.

Cheers

Constructing phylogenetic trees by using whole genome sequence alignments is conceptually a wrong approach. No single phylogenetic parameter can deal with the highly variable evolutionary signals within a genome. For accurate phylogenetic analysis you need to select sub-segments of genome with identical evolutionary properties and than apply parameter by using knowledge base approach.

Dear Azeem:

I would suggest the program RDP3 (Martin D, Rybicki E (2000) RDP: detection of recombination amongst aligned sequences. Bioinformatics 16: 562–563.) which Dr. Martin modified for our uses to be able to do whole bacterial genome alignments. We published this modified method together (Hiller, N.L., Ahmed, A., Powell, E., Eutsey, R.E., Earl, J., Martin, D., Janto, B., Hogg, J.S., Boissy, R., Barbadora, K., Post, J.C., *Hu, F.Z., and *Ehrlich, G.D. Generation of Genic Diversity among Streptococcus pneumoniae Strains via Horizontal Gene Transfer during a Chronic Polyclonal Pediatric Infection. PLoS Pathogens 6(9): e1001108, 2010).. Since it works for bacterial genomes I don't think you'll have trouble with viruses

Thank you everyone for your replies!!

TOPALi didn't work. After running upto 50% for nucleotide substitution model, it gave error. RDP 4.16 beta also ran for 20hrs and then it stopped working. For now I have divided my data into 40 sequences out of 100. I then performed alignment with MAFFT server. From this alignment, I estimated nucleotide substitution model using MEGA5 (Results are attached). Based on these results, please tell me which model is the best and how to give its input in MEGA5 maximum likelihood tree construction dialog box

CIPRES Science Gateway (http://www.phylo.org/portal2/login!input.action), as pointed out by Diogo, is very good. Alternatively, you also have University of Oslo Bioportal (www.bioportal.uio.no). I don't know if the will "like" a big dataset as yours though.

you can upload your data to a server or mainframe in your university or institute nearby.

how many CPUs in your computer?

Amir Abbasi is right, select some genes that can give a phylogenetic relationship, using the whole genome can give disorted results. For bacteria, using genome "signatures" like ANI, TEM, etc. is a better approach, if interested check here http://www.imedea.uib.es/jspecies/

Dear Azeem,

I would suggest GreenButton. GreenButton is a plugin that builds the power of a Supercomputer into Geneious. It allows you to complete intensive analyses in a fraction of the time it would normally take you – and it's just as easy as running the analyses on your desktop computer.

GreenButton allows you to run your analyses on Microsoft's Windows Azure platform. The job is submitted to Windows Azure right from within Geneious via the GreenButton plugin. From there it is distributed across several cores where it is processed. When the job is complete, the results will be automatically delivered to your copy of Geneious, and you can also download the files from the Green Button website.

GreenButton uses a pre-pay system and charges US 18¢ per core hour.

http://www.geneious.com/web/geneious/greenbutton

GreenButton website: http://www.greenbutton.com/

Geneious website: http://www.geneious.com/

Dear Azeem,

The best model is the first in the list in the MEGA5 output, almost always. You need to take into account the model with the lowest BIC, the lowest AIC and with the highest ML. And, i think that you can use the selected model in your analysis just clicking one tag in the corresponding window.

Hi, you can use RAML or GARLI, or if you want... use Phylogeny.fr plattaform for execute extensive analyses. Greetings

Advice on which methods to use, without knowledge of what the issues are with the alignment problem, is likely to be wrong. Aligning hundreds of genomes of 36kb length that are all very similar to each other is a completely different problem than aligning genomes that have gene rearrangements or other issues. An alignment tool that produces a result quickly may be useless if it produces a bad result. Conversely a tool that in theory will give an "optimal alignment" will be useless if it takes six years on a supercomputer to complete the task. The best tool to use is nearly always some compromise. Which family/species of viruses are you attempting to align?

You could try Mulan at http://mulan.dcode.org/

Dear Azeem,

Have you ever tried MAFFT? It is available on line and uses CPUs from a good data server.

Use this protocole.... I hope you like it.