etetoolkit

Source: https://github.com/aipoch/medical-research-skills

When to Use

• Preprocess phylogenetic trees: convert formats (Newick/NHX/PhyloXML), reroot (midpoint/outgroup), prune taxa, and resolve polytomies before downstream analyses.
• Detect evolutionary events in gene trees: infer duplication vs. speciation events and derive ortholog/paralog relationships for phylogenomics.
• Annotate trees with taxonomy: map species names to NCBI TaxIDs, retrieve lineages/ranks, and build minimal taxonomy topologies connecting a set of taxa.
• Generate publication-quality visualizations: render trees to PDF/SVG/PNG with custom styles, support-based coloring, and node “faces” (labels, shapes, heatmaps).
• Compare alternative topologies: quantify differences between trees using Robinson–Foulds (RF) distance and partition/bipartition analysis.

Key Features

• Tree I/O and manipulation
  • Read/write: Newick, NHX, PhyloXML, NeXML
  • Traversals: preorder, postorder, levelorder
  • Operations: prune, reroot, collapse, resolve polytomies
  • Metrics: branch/topological distances, RF distance
• Phylogenetic (gene tree) analysis
  • Alignment association (FASTA/Phylip)
  • Species name extraction from gene IDs
  • Duplication/speciation detection (e.g., species overlap / reconciliation-style workflows)
  • Orthology/paralogy extraction and gene-family splitting
• NCBI taxonomy integration
  • Auto-download + local cache of taxonomy DB
  • TaxID ↔ scientific name translation
  • Lineage/rank retrieval and taxonomy-based topology building
  • Tree annotation with taxonomic metadata
• Visualization
  • Rectangular/circular layouts, GUI exploration
  • NodeStyle/TreeStyle customization
  • Faces (text, shapes, charts/heatmaps) and layout functions
  • Export to PDF/SVG/PNG
• Clustering support
  • ClusterTree for dendrograms linked to numeric matrices
  • Cluster quality metrics (e.g., silhouette, Dunn index)
  • Heatmap + tree combined views

Dependencies

• ete3 (recommended: >=3.1.0)
• Optional GUI/rendering dependencies (platform-specific):
  • PyQt5 (e.g., >=5.15)
  • Qt SVG support (often packaged as python3-pyqt5.qtsvg on Debian/Ubuntu)

Example Usage

The following example is designed to be runnable end-to-end (it uses an in-memory Newick string and does not require external files).

# pip install ete3

from ete3 import Tree, TreeStyle, NodeStyle

# 1) Load a tree (Newick)
nw = "((A:0.1,B:0.2)90:0.3,(C:0.2,D:0.4)70:0.1);"
t = Tree(nw, format=1)

# 2) Basic stats
print("Leaves:", len(t))
print("Total nodes:", sum(1 for _ in t.traverse()))

# 3) Midpoint rooting
mid = t.get_midpoint_outgroup()
t.set_outgroup(mid)

# 4) Prune to taxa of interest (preserve branch lengths)
t.prune(["A", "C", "D"], preserve_branch_length=True)

# 5) Style nodes (color internal nodes by support)
ts = TreeStyle()
ts.show_leaf_name = True
ts.show_branch_support = True

for n in t.traverse():
    st = NodeStyle()
    if n.is_leaf():
        st["fgcolor"] = "blue"
        st["size"] = 8
    else:
        # ETE stores internal support in n.support when present
        st["fgcolor"] = "darkgreen" if getattr(n, "support", 0) >= 80 else "red"
        st["size"] = 5
    n.set_style(st)

# 6) Render (PDF/SVG/PNG supported depending on your environment)
t.render("example_tree.pdf", tree_style=ts)
print("Wrote: example_tree.pdf")

Implementation Details

Tree parsing formats (Newick “format” codes)

ETE uses a format integer to control how node attributes are interpreted when reading/writing Newick. Common patterns:

• format=0: flexible default (often includes branch lengths)
• format=1: includes internal node names
• format=2: includes support/bootstrap values
• format=5: internal node names + branch lengths
• format=8: name + distance + support (maximal common usage)
• format=9: leaf names only
• format=100: topology only

Example:

from ete3 import Tree

t = Tree("tree.nw", format=1)
t.write(outfile="out.nw", format=5)

NHX feature preservation

NHX is used to store custom per-node features. When writing, specify which features to serialize:

t.write(outfile="tree.nhx", features=["taxid", "habitat", "lineage"])

Rerooting and pruning behavior

• Midpoint rooting uses get_midpoint_outgroup() to select an outgroup that balances path lengths.
• Pruning should typically use preserve_branch_length=True to avoid distorting distances in phylogenetic contexts.

Evolutionary event detection (gene trees)

For gene trees, PhyloTree supports event labeling on internal nodes (commonly:

• evoltype == "D" for duplication
• evoltype == "S" for speciation)

A typical workflow is:

1. Load a gene tree (optionally with an alignment).
2. Provide a species naming function to map gene IDs → species.
3. Run descendant event detection.
4. Extract ortholog groups (speciation subtrees) or query ortholog/paralog sets from events.

Tree comparison (Robinson–Foulds)

Tree.robinson_foulds(other_tree) returns:

• rf: RF distance (number of differing bipartitions)
• max_rf: maximum possible RF given shared leaves
• plus shared leaves and partition sets for deeper inspection

Normalized RF is typically computed as rf / max_rf (when max_rf > 0).