10 October 2018, Wednesday. For last week’s lab HW, I had to make a folder with three of my samples (CO-01,02 and 03), 25 sequences of the COI gene for Actinopterygii and at least one COI gene for Chondrichthyes (outgroup). To find those sequences, I had to log into Geneious and click NCBI nucleotide section and then type in COI Actinopterygii or Chondrichtyes on the search bar. I put all 29 samples into one folder and made an nucleotide alignment sequence.

Phylogenetic Inference

First thing that I did was to clean up my alignment to begin and end at the same point. I just clicked ‘Allow editing’ and select the bases and deleted them. Looking at the first 20 columns of my alignment, I found 12 polymorphic nucleotides. Next I downloaded JModelTest2 from Canvas. I had to download Java to open up the file jmodeltest2 because my computer was not able to open up the file. After that, I went back to Geneious and export my edited alignment in Phylip format (relaxed) by choosing Export, selected documents’, ‘Phylip’ file type and relaxed Phylip file at the end. And using JModelTest 2, I opened up the exported alignment. Next I chose ‘Analysis > Compute likelihood scores’ from the menu. My alignment took about 5 minutes for the program to look for the likelihood for each of the models to be the best fit. After the calculations were done, I looked up AIC and BIC values (red, top scoring model) and found out that the values were the same with HKY +G.

Bayesian Inference

Back in Geneious, I clicked Tools, Plugins and install the Bayes plugin. Then selected my alignment, right click and chose Tree then MrBayes. Changed the substitution Model to HKY, Rate Variation to Gamma, Outgroup to shark, Chain length to 10,100, Subsample frequency to 99, Burn-in Length to 100 and left the heated chains at 4. I then run the analysis and didn’t took that long. Double-click on new ‘Posterior output’ to open up the result and click on the ‘parameter estimates tab’. I got really bad graph as expected and also got bad Trace as expected because the run time was very fast. This time I changed chain length to 100,000 and bur- in length to 10000. It took about 10 minutes and I got a pretty good graph.

Maximum Likelihood

This time, I went to Tools, Plugins and installed RAxML. This time, instead of choosing Bayesian, I chose ‘Rapid bootstrap with rapid hill climbing’. Once RAxML was done, I clicked ‘RAxML bootstrapping trees and right clicked to choose Tree and consensus tree builder. After that I chose ‘Create consensus tree’ and ‘Support threshold’ of 50%, and then ‘ok’. I had a decent tree and for my report I had to take two sequences out because the branch was too long. The other steps didn’t work so i had to stop at PHYML part.