A tricky question. I think RAxML is prone to artifacts, MrBayes requires belief in your priors, and both PhyML and IQTree are quite sensitive to model selection. So, once you are done with selecting a reasonable model, perhaps your best path would be to show the IQTree, carefully point out any (minor) discrepancies with PhyML, and provide support from other methods for every node. Hope ths helps!

Thank you! This is how I've done consensus trees previously, so I will do that this time as well.

So, if RAxML is prone to artefacts, but the tree topology agrees with the other topologies pretty well, is it safe to say I have a pretty solid tree? I'm noticing that the patterns I'm getting are pretty consistent. Again, the deeper nodes are not so strong no matter which method I use simply because these proteins are not highly similar in places (the Wnt domain is a series of conserved cysteine positions and the in-betweens can vary quite a bit). Thanks again, this is very helpful since I am no longer in a research lab, but am trying to wrap up the re-writing of a manuscript.