Which Capsaspora T-box protein is Brachyury (CoBra)?

This is our own analysis (2026-10-01), not published data. Reproduce it with:

uv run python projects/ORIGINS_OF_MULTICELLULARITY/capsaspora-tbox/tbox_assign.py

The raw output is in tbox_assign_output.txt. All sequences and domain
coordinates are fetched live from UniProt.

Problem. PMID:24043797 characterised Capsaspora Brachyury (CoBra) but
gives no locus ID. UniProt has three Capsaspora T-box proteins, all with
automatic names:

Two tests.
1. Best match. We scored each T-box domain against 12 human T-box
domains by global alignment (BLOSUM62).
2. Residue marker. We found the residue aligned to Lys149 of Xenopus
Brachyury (P24781). The paper uses this position as a marker: Lys in
metazoan Brachyury, Asn in other T-box classes, and Arg in CoBra.

Results.

Assignment.
- A0A0D2VUC6 (CAOG_005512) is CoBra. It is the best T-subfamily match,
and it has the arginine the paper reports for CoBra at the Lys149
position.
- A0A0D2WSA5 scores closest to human TBX2. We do not map it to the paper's
second gene. The paper calls that gene "CoTbx3" but places it in a new Tbx7
class of non-Brachyury holozoan T-box genes ("a previously undescribed class
of T-boxes (Tbx7), which includes the remaining (non-Brachyury) filasterean
and ichthyosporean T-box genes"). Human T-box domains cannot represent a
class with no human member. Corrected 2026-10-01; an earlier version called
A0A0D2WSA5 "consistent with the paper's CoTbx3".
- Caveats. The assignment rests on similarity plus one marker residue, not
a phylogeny. The paper also mentions a Capsaspora T-box gene with two
T-box domains; none of the three UniProt entries has two annotated T-box
domains, so the gene models may differ from the paper's.