Structural phylogenetics: putting protein structure into phylogenetic inference

Structural phylogenetics is an emerging area that attempts to add information from protein structure to traditional amino-acid-based phylogenies of proteins. As protein structure is typically more conserved than sequence, this approach is likely to be useful for phylogenetic problems where amino acid sequence similarity is highly decayed due to long time periods or rapid evolution (e.g. viruses). In such situations, alignment and accurate phylogenetic inference become very difficult.

Structural phylogenetics has been the subject of recent efforts, with several different methods proposed. The advent of AI/LLM-assisted structural prediction algorithms like AlphaFold has made it feasible to apply these methods without being limited to only experimentally solved Protein Data Bank (PDB) structures.

This course will review various proposed approaches and their assumptions, and guide students to implement and compare several of them. We will put the most focus on Maximum Likelihood methods, where standard amino acid (AA) data is combined with “3Di” (3-dimensional) structure characters derived from the FoldSeek program in a partitioned analysis. As part of the workshop, therefore, students will learn the bioinformatics needed to get sequences, get AlphaFold structures, generate 3Di characters, do structural alignments, estimate phylogenies, and visualize and interpret results. We will conclude by reviewing the strengths and weaknesses of current methods and highlighting the many open research questions in structural phylogenetics.

Places are limited to 15 participants.