“…Addressing the widely discussed issue of background noise (Street et al, 2017 ; Sanderson et al, 2018 ; Charalampous et al, 2019 ; Yang et al, 2019 ; Chan et al, 2020 ), we found herein that setting a cut-off of 1% classified reads improved the accuracy of results, in line with many published studies (Sanderson et al, 2018 ; Charalampous et al, 2019 ; Yang et al, 2019 ; Chan et al, 2020 ). Nonetheless, using this cut-off in sample 31, collected from pleuropneumonia, led to categorising some bacteria as background noise while these are well-known oral flora bacteria (Nakazawa et al, 2002 ; Heintz et al, 2019 ). Conversely, the analysis of sample 22 (aortic valve in a context of infective endocarditis) concluded with the notable presence of six bacteria including environmental and oral flora bacteria, questioning their relevance as etiological agents.…”