Summary

In the work presented, we develop culture-free laboratory workflows and tailored bioinformatic pipelines to obtain high quality genomes from sputa. We include samples from two different settings, Mozambique and Georgia and reanalyse published data to ensure generalisation. Key findings include: Culture accurately reflects sputum genetic diversity in all settings evaluated; our benchmarking of culture-free protocols reveals key sources of false genetic diversity. After discarding sources of false variation, differences between sputum-culture samples almost disappeared in all datasets evaluated; our data suggest that the limit of detection of culture-free sequencing is still far from a desired point-of-care test. Overall our results evidence that the large body of work in tuberculosis research, based on culture sequencing, accurately reflects the diversity present in sputum. We also provide insights into the future role of culture-free sequencing in diagnostics and anticipate that our results will be of utmost interest to the tuberculosis community and may even provide lessons (and precedents) for similar issues in other pathogens.