Phylogenetic Tree

Sequences for this tree

Download:

For iTOL: upload the Newick file, then drag the two .txt files onto the displayed tree to add the annotation tracks.

Tree + Domain Architecture

Download:

ETE4 Interactive Explorer

Clade actions ? Pick a branch, then choose what to do with it. To pick one, right-click any branch in the ETE4 explorer above and choose “Use branch for divergence” — it is highlighted and appears here. You can select more than one.

Build a tree first. Fill in the form on the left and press Run Tree Pipeline. These actions work on a branch of the tree you build. To pick branches by right-clicking, choose the ETE4 server viewer.

No branch selected — actions below will use the whole tree.

Action

Working…

MSA divergence analysis ? Which positions in this family's alignment behave differently between subfamilies? Three scores, computed from the alignment this tree was built from:

Conservation (JSD) — the position is unusually fixed across the whole family. Needs no groups.
Group specificity (GroupSim) — fixed within each subfamily but different between them: the signature of a position that determines what a subfamily does.
Diagnostic power (MI) — how well the residue tells you which subfamily a sequence belongs to. 1.00 means with certainty.

All three down-weight near-identical sequences, so a clade sequenced thirty times does not count thirty times.

Build a tree first. Then the three scores are computed from the alignment it was built from.

Compare Automatic cuts the tree at half its depth — a quick look before you pick anything.

Rank sites by

How many sites to show

Number positions relative to ? Alignment columns are only meaningful against this alignment. Pick a sequence here and every site also gets its residue number in that protein — the number you can look up in its UniProt entry or point at in a structure.

The list fills in after the first scoring run, because the sequence names come back with the scores. Changing it re-scores immediately; the scores themselves do not depend on it. Score once, then pick a sequence to get UniProt residue numbers.

Gap threshold ? Group specificity skips any column where either group is more than this fraction gaps — comparing residues across a mostly empty column invents differences that are not there.

The cost is that one ragged or partial sequence can push most of the alignment past the limit and leave you reading a narrow window without noticing. The summary line below always says how many columns were actually scored. If that number is small, either raise this threshold or drop the short sequence and re-align the clade.

0.1 is the default. 1.0 turns the filter off entirely. Higher scores more columns, on thinner evidence.

Exclude sequences ? Leave these out of the scoring. The tree is untouched — only what the scores are computed from changes.

This is what to reach for when the summary line says only a fraction of the columns were scored. A partial or fragmentary sequence is gappy everywhere the rest of the family is not, so it fails the gap threshold column after column and takes most of the alignment with it. Dropping one such sequence can take a run from 86 scored columns to 347.

Full leaf names, separated by spaces or commas. Anything that is not in the alignment is reported back rather than ignored. Drop a fragment without rebuilding the tree.