This dataset includes FastQC reports for raw data (Supplementary Material 1), FastQC reports for trimmed data (Supplementary Material 2), Fastp reports on the impact of data quality filtering (Supplementary Material 3), and mapDamage plots of the human DNA sequences recovered from the datasets (Supplementary Material 4).

Detailed methods may be found in the associated publication. The dataset was collected from archaeological dental calculus and associated bone samples. DNA was extracted within dedicated biomolecular clean labs. After crushing to a powder, samples were pre-digested for 5 minutes with 1 mL of 0.5M EDTA to remove possible surface contamination. This pre-digestion supernatant was removed, and a further 1.1 mL of 0.5M EDTA was added and rotated at room temperature for seven days to fully demineralize. For the majority of samples, DNA was extracted from the dental calculus and bone samples using a protocol based on Dabney et al. (2013). All DNA extracts were quantified via Qubit® 2.0 Fluorometer using a High-Sensitivity DNA Assay. For each DNA extract, double-stranded whole genome shotgun Illumina libraries were prepared using a protocol based on Meyer and Kircher (2010). The dental calculus libraries were pooled in equimolar concentration and subjected to paired-end sequencing on multiple HiSeq2500 lanes at the Wellcome Trust Sanger Institute (WTSI) or on a NextSeq platform, PE 150 + 150 bp Integrated Microbiome Resource (IMR) at Dalhousie.

FastQC v0.11.9 was used to assess raw digital data quality. FastQC is a quality control tool for raw sequencing data that provides a modular collection of analyses used to gain insight into any flaws in the data before performing further analysis. The preprocessing programmer Fastp v0.23.2 was then utilised with default parameters. Fastp is a tool used to filter and trim poor-quality reads, cut adapters, repair mismatched base pairs, and produce overall quality. It also provides results that include both pre- and post-filtering data, allowing for a direct comparison of the filtering impact. Centrifuge v1.0.3 was used, with default parameters, to assign taxonomic labels by mapping sequences against the human genome, prokaryotic genomes, and viral genomes, including 106 SARS-CoV-2 complete genomes. Human reads from Centrifuge outputs were retained, and seqtk ‘subseq’ was used to convert them into fastq files that were then mapped to the human genome (hg38) (NCBI 2013) using BWA mem v0.7.17 . SAMtools v1.12  (-view -rmdup -flagstat -sort -index) was then utilised for alignment formatting and was sorted into BAM files that were run through mapDamage2 v2.2.2 with default parameters.