Researchers have developed a novel tree-conditioned edit-flow model for ancestral sequence reconstruction (ASR), aiming to infer extinct protein sequences. This model reconstructs ancestors by using paired bidirectional edit trajectories that converge on a common ancestral state, accommodating insertions, deletions, and substitutions. While it did not outperform classical methods on a benchmark with only context-independent substitutions, it showed strong performance on natural homologous sequences with complex evolutionary changes. AI
Summary written by gemini-2.5-flash-lite from 1 source. How we write summaries →
IMPACT Introduces a new method for inferring protein evolution, potentially improving biological sequence analysis.
RANK_REASON This is a research paper published on arXiv detailing a new model for ancestral sequence reconstruction. [lever_c_demoted from research: ic=1 ai=1.0]