The Bone or the Genome, a Comparative Study of Hypothesis-driven and Blind Screening Approaches in Ancient Pathogen Genomics

Abstract

The rapid growth of high-throughput sequencing datasets from archaeological remains has expanded ancient pathogen genomics beyond single, targeted investigations. Pathogen work can now begin either from context-first sampling guided by archaeological and osteological hypotheses or from genome-first screening of existing shotgun libraries. This thesis compares these entry points using two paired case studies processed under the same metagenomic, mapping-based, and authentication framework. The context-first case examines two adult crania curated at the Museo de La Plata and attributed to Gran Canaria, each displaying craniofacial changes consistent with rhinomaxillary syndrome. Shotgun sequencing and screening yielded low but non-zero mapping signal to Mycobacterium leprae across several extracts, with stronger apparent signal in dense elements and lesion-proximal material. However, stringent specificity filtering (BLASTN followed by MEGAN) reduced retained M. leprae reads to single digits in most extracts, and breadth of coverage remained extremely limited. The resulting information content was insufficient to support stable phylogenetic placement or clade-level assignment, preventing inference about introduction routes to the Canary Islands. The genome-first case analyses demultiplexed shotgun data from a radiocarbon- dated infant petrous bone from the Uspallata Valley (Mendoza, Argentina, 607 BP) with no pathognomonic skeletal indicators of infection. After screening, mapping, and specificity validation, a coherent set of Treponema pallidum reads supported phylogenetic placement. Using EPA-ng on a fixed reference tree, the infant query sequence placed basal to the published Argentine DFU001 lineage, providing an interpretable position within regional treponemal diversity despite low coverage. Together, these case studies show how starting conditions shape what ancient pathogen workflows can support.