Analysis scripts for independent validation of TERT–UG17/DRT10 (Class 2) evolutionary proximity

This repository contains the complete Python and R analysis scripts used toindependently validate the evolutionary proximity between eukaryotic telomerasereverse transcriptase (TERT) and bacterial UG17/DRT10 (Class 2) defense reversetranscriptases, using public data only.The pipeline comprises four stages:(1) RT discovery and classification — hmmsearch of the AlphaFold Database v6  proteome with 35 class-specific HMM profiles (UG subgroups, AbiA/AbiK/AbiP2,  TERT) plus Pfam/CDD/TIGRFAM RT profiles, followed by best-hit class  assignment, cross-hit exclusion (RdRP, maturase K) and CD-HIT deduplication;(2) high-quality core set construction — multi-evidence filtering and per-class  core-subcluster extraction, yielding 1,480 core RT proteins;(3) phylogenetic inference — all-vs-all structural alignment with Reseek,  custom NumPy-vectorized neighbor-joining tree construction, and  maximum-likelihood validation with IQ-TREE 2 (LG+G4, 1,000 UFBoot);(4) figure generation — circular tree visualization with R/ggtree, inter-class  distance heatmaps, TERT nearest-neighbor bar plots, and host-taxonomy  composition charts.Key result: TERT is the nearest neighbor of UG17 (DRT10) both at the sequencelevel (IQ-TREE patristic distance 3.339) and at the RT-domain structure level(median distance 0.293; 0.298 after excluding eukaryotic Class2 members),reproducing the conclusion of Tang et al. 2025 (bioRxiv 2025.10.16.682844).All third-party tools (HMMER, CD-HIT, Reseek, Foldseek, MAFFT, trimAl,IQ-TREE 2, ggtree) are cited in the accompanying manuscript; the depositedcode is the custom analysis and figure-generation logic written by the authors.