I have thousands influenza sequences and tried so many times. But failed. Is there any detailed demo from the begaining of sequence aligenment to phylogenetic tree building? I hope someone could show me the detailed protocol of building ML phylogenetic tree by RAxML-HPC2 on XSEDE (8.2.8) on cipres.
Hello Li,
Here is the help page for RAxML on cipres:
http://www.phylo.org/index.php/tools/raxmlhpc8_xsede.html
But here is a quick tutorial.
First of all I suggest you run jModelTest to test the for the best model and to see if GTR (RAxML default) is really the best model. I also suggest building a phylogeny with a Bayesian program as well (MrBayes is on Cipres)
1) Click the data folder in your project and upload a phylip file (i.e. .phy or .phylip) by selecting the upload data button and following the steps.
2) Click the tasks folder in the same project and click the select input data button to select your data.
3) Click the select tool button and select RAxML-HPC2 on XSEDE (8.2.8).
4) Click the 36 parameter set button and adjust the parameters to fit the analysis you wish to preform. One aspect I would suggest changing is to is under the configure the analyses section. Under the select the analyses type pull down menu change from default to Rapid Bootstrap Analyses (-f a). Without this you will not get a bootstrap tree which is what most people want. You also will need to select Rapid Bootstrapping under the configure bootstrapping section. You can also change the number of bootstraps below, but the default is 100 which is pretty normal (although some journals want you to run 1000, but this is rare). If you want to run a different analyses then consult the user manually for RAxML which is in the link above. Also make sure you set an appropriate amount of time, a quarter of an hour is usually more than enough for single gene or other small datasets. If you are using NGS data I would suggest setting this higher. You can also just change it to one hour and they will only charge you for the time used and you do not change much in the queue.
5) Give your task a name and run.
6) Once complete click view output and download one of the RAxML_bipartitions files. There is one that is RAxML_bipartitions.[whatever name you gave, default name is result] and one that is RAxML_bipartitionsBranchLabels.[what ever name you gave]. Different programs for visualizing your tree will require one format or the other, FigTree for example will open the first but with throw up an error with the latter.
If you do not have the files listed above then an error has occurred and you should view the STDOUT to see what error killed the program.
Hope this helps.
Best Regards
Hello Li,
Here is the help page for RAxML on cipres:
http://www.phylo.org/index.php/tools/raxmlhpc8_xsede.html
But here is a quick tutorial.
First of all I suggest you run jModelTest to test the for the best model and to see if GTR (RAxML default) is really the best model. I also suggest building a phylogeny with a Bayesian program as well (MrBayes is on Cipres)
1) Click the data folder in your project and upload a phylip file (i.e. .phy or .phylip) by selecting the upload data button and following the steps.
2) Click the tasks folder in the same project and click the select input data button to select your data.
3) Click the select tool button and select RAxML-HPC2 on XSEDE (8.2.8).
4) Click the 36 parameter set button and adjust the parameters to fit the analysis you wish to preform. One aspect I would suggest changing is to is under the configure the analyses section. Under the select the analyses type pull down menu change from default to Rapid Bootstrap Analyses (-f a). Without this you will not get a bootstrap tree which is what most people want. You also will need to select Rapid Bootstrapping under the configure bootstrapping section. You can also change the number of bootstraps below, but the default is 100 which is pretty normal (although some journals want you to run 1000, but this is rare). If you want to run a different analyses then consult the user manually for RAxML which is in the link above. Also make sure you set an appropriate amount of time, a quarter of an hour is usually more than enough for single gene or other small datasets. If you are using NGS data I would suggest setting this higher. You can also just change it to one hour and they will only charge you for the time used and you do not change much in the queue.
5) Give your task a name and run.
6) Once complete click view output and download one of the RAxML_bipartitions files. There is one that is RAxML_bipartitions.[whatever name you gave, default name is result] and one that is RAxML_bipartitionsBranchLabels.[what ever name you gave]. Different programs for visualizing your tree will require one format or the other, FigTree for example will open the first but with throw up an error with the latter.
If you do not have the files listed above then an error has occurred and you should view the STDOUT to see what error killed the program.
Hope this helps.
Best Regards
I agree with most of what Mr Max described.
I would just like to add on sequence alignment:
Before you test your data for the best fit model, align your sequences: there are many online programs for MSA (I have been recently using MAFFT).
After you have aligned your sequences using the program, make sure you check the alignment manually for any sort of faults in the alignment (sometimes due to erroneous global alignment) and edit as necessary--using editing tools in common phylogenetic software e.g. MEGA).
Then, export and save your data matrix in the required format for further analysis.
Best regards.
- Badges
- Science topic
- Similar topics
- Biological Science
- Evolutionary Biology
- Phylogenetics