Replace RAxML tree building with IQ-Tree

[j23414]

Context

The West Nile Virus (WNV) pipeline currently uses RAxML for tree building, which is taking up a significant portion of the runtime (about 4 out of 6 hours).

Based on some online reading about RAXML generally runs much slower than IQ-Tree, I checked in with @DOH-LMT2303 and @jameshadfield if we could switch to IQ-Tree or if there was a technical reason why RAXML was selected over IQ-Tree for WNV.

• @DOH-LMT2303 investigated the literature and found that both RAxML and IQ-TREE have been used in WNV studies. Based on this, she concluded that we could probably use any tool that produces a maximum likelihood phylogeny and get similar results.
• We noted that Grubaugh lab has switched to using IQ-TREE as the default for their WNV-global build.
• @jameshadfield confirmed that RAxML was used in the original build because it was the standard at the time. He stated that if he were to create the build today, he would use IQ-TREE.

There doesn't appear to be any WNV-specific reason for using RAxML over IQ-TREE, but let me know if if I'm mistaken. I'm going to explore swapping in IQ-Tree and see if we get any speedups

Description

Swap in IQ-Tree and determine if we get any speedups to further avoid the 6hr GitHub Action runtime limit.

Examples

Possible solution

[j23414]

Switching to IQ-Tree (view tree here) reduced runtime drastically:

• Before: https://github.com/nextstrain/WNV/actions/runs/11920429171 (4:57:42 ~ nearly 5 hours for the tree building benchmark)
• After: https://github.com/nextstrain/WNV/actions/runs/11938827520 (0:04:13 ~ 4 minutes for the tree building benchmark)