A Pipeline for Generating Datasets of Three-Dimensional Tertiary Interaction Characters for Model-Based Structural...

Tertiary-interaction characters, termed "3Di" characters, were invented for deep structural homology search with the FoldSeek program. However, they may also be used as phylogenetic characters in model-based phylogenetic inference, just as in the Maximum Likelihood program IQ-TREE. However, to conduct such an analysis, every input amino acid (AA) sequence must be linked to a protein structure prediction file, which then must be converted to 3Di characters, and then both AA and 3Di characters need to be assembled, aligned, and joined in a partitioned data file for phylogenetic analysis. To make this process practical for medium to large datasets, we constructed a pipeline with R scripts to automate these steps. Here we present example scripts and explain the R functions used to construct a joint AA+3Di alignment file.