Per my comment to the question, here is an answer to the same question asked on ResearchGate:
Whole genome sequencing of human tissue samples often results in reads aligning to bacterial references, and this is actually a method used in the diagnoses of infectious diseases.
Identification of low abundance microbiome in clinical samples using whole genome sequencing
However, because pathogen DNA may be present at a much lower abundance than host DNA, it may be difficult to distinguish true infections from contamination and false positives. Steven Salzberg addresses these concerns and offers a computational solution in a recent publication:
KrakenUniq: confident and fast metagenomics classification using unique k-mer counts
"Usually, the vast majority of the reads match (typically 95–99%) the host, and sometimes fewer than 100 reads out of many millions of reads are matched to the target species. Common skin bacteria from the patient or lab personnel and other contamination from sample collection or preparation can easily generate a similar number of reads, and thus mask the signal from the pathogen."
To expand on this answer, it is important to realize that accurate identification of low-abundance organisms in metagenomic reads requires very good reference genomes. A separate paper from the Salzberg group discusses a startling discovery: many high-copy repeats in the human genome have been incorrectly annotated as bacterial proteins in the NCBI RefSeq database.
Human contamination in bacterial genomes has created thousands of spurious proteins
This suggests that researchers should be careful when attempting to infer the presence of low-abundance bacteria or phage in human tissue sequencing data, particularly when the presence of those organisms is not corroborated by other analyses.
