I am working on construction of phylogeny based on Mrbayes and RAxML. I used the Prottest to get the best model for ML (FLU+I+G+F) and BIC (FLU+G+F). I tried FLU for RAxML,but when I used RAxML I could not find the FLU model, so instead i used the prset aamodelpr=fixed(mix) parameter and the third ranked model from ProtTest(GTT+G+F) with preset aamodelpr= fix(jone). But the results are totally different from RAxML results, which is completely not what i expected. So can anyone give me some suggestion? As the results showed the most significant bias is that one of my specials owns longer branch than its closest species.