NGS data corresponding to 24 ancient Native Brazilians - metagenomic reads unmapped to the human reference

We kindly ask users to observe the Fort Lauderdale Agreement principles, which entitles the data producers to make the first presentation and publish the first analysis of specific microbial genomes/genes present in that dataset. For some related results see: Cruz Dávalos, D. I. et al. Indigenous peoples in eastern Brazil: Insights from 19th century genomes and metagenomes. 2022.01.27.477466 Preprint at https://doi.org/10.1101/2022.01.27.477466 (2022). This non-human part of this study is made available in a companion study [DOI: https://doi.org/10.1038/s41598-023-40246-x] - see below for a description. Hence, this data can be used freely for compositional and community-oriented microbiome analyses as performed here [DOI: https://doi.org/10.1038/s41598-023-40246-x]. Study description: Reconstructing the history - such as the place of birth and death - of an individual sample is a fundamental goal in ancient DNA (aDNA) studies. However, knowing the place of death can be particularly challenging when samples come from museum collections with incomplete or erroneous archives. While analyses of human DNA and isotope data can inform us about the ancestry of an individual and provide clues about where the person lived, they cannot specifically trace the place of death. Moreover, while ancient human DNA can be retrieved, a large fraction of the sequenced molecules in ancient DNA studies derive from exogenous DNA. This DNA - which is usually discarded in aDNA analyses - is constituted mostly by microbial DNA from soil-dwelling microorganisms that have colonized the buried remains post-mortem. In this study, we hypothesize that remains of individuals buried in the same or close geographic areas, exposed to similar microbial communities, could harbor more similar metagenomes. We propose to use metagenomic data from ancient samples' shotgun sequencing to locate the place of death of a given individual which can also help to solve cases of sample mislabeling. We used a k-mer-based approach to compute similarity scores between metagenomic samples from different locations and propose a method based on dimensionality reduction and logistic regression to assign a geographical origin to target samples. We apply our method to several public datasets and observe that samples from closer geographic locations tend to show higher similarities in their metagenomes compared to samples of different origin, allowing good geographical predictions of test samples. Moreover, we observe that the genus Streptomyces commonly infiltrates ancient remains and represents a valuable biomarker to trace the samples' geographic origin. Our results provide a proof of concept and show how metagenomic data can also be used to shed light on the place of origin of ancient samples.