Lineage-specific selection signals in the Growth arrest-specific protein 8 (GAS8) domain protein of Trypanosoma melophagium

The Growth arrest-specific protein 8 (GAS8) domain-containing gene family (Orthogroup 6127at5690) is a highly conserved, single-copy orthologous group in the Trypanosoma taxa retained for codon-based evolutionary analysis (10 validated species). This family encodes a microtubule-associated protein implicated in cytoskeletal organization and flagellar motility. We conducted a genus-wide evolutionary analysis of the GAS8-domain protein, focusing on the Trypanosoma melophagium ortholog LSM04 004638. Maximum likelihood phylogenetic reconstruction, branch-specific and site-level codon-based selection analyses, ancestral sequence reconstruction, and Rosetta-based energetic estimation were performed. Despite strong purifying constraint consistent with essential cytoskeletal function, branch-level aBSREL detected a statistically significant signal compatible with episodic diversifying selection along the T. melophagium lineage (aBSREL: LRT = 6.72, Holm-corrected p = 0.0494). Site-level analyses identified two significant MEME codon sites (2 and 109; p = 0.00598 and p = 0.00916), corresponding to derived non-synonymous substitutions relative to the reconstructed ancestor (P2G and V106L); a third site (267) was suggestive but not significant (p = 0.08596). Branch-site codeml did not reach statistical significance (LRT = 1.7428, p = 0.1868, BH-FDR = 0.9339), although BEB highlighted site 2 (PP = 0.983) under the alternative model. These results are consistent with, but do not by themselves prove, limited lineage-specific adaptive refinement in an otherwise highly constrained cytoskeletal regulator; the signal should be regarded as method-dependent and statistically modest based on codeml branch-site test. A RELAX sensitivity test found no evidence that the focal-lineage signal is driven by relaxed constraint (K = 1.12, p = 0.697). Rosetta-based energetic comparisons across single- and double-reversion models consistently yielded positive [Formula: see text] values, but structural interpretation remains model-dependent and should be treated as hypothesis-generating rather than as direct evidence for altered function in vivo.