Clinical sample selection

All IS6110-RFLP and Spoligo patterns of the M. tuberculosis isolates from 2004 to 2020 were loaded in the Bionumerics database (v7.6, Applied Maths, Kortrijk, Belgium). Among the 2553 genotyped isolates, we selected two or three isolates belonging to each of the IS6110-clusters involving 10 or more cases of TB: CLS 2 (one isolate from 2007 and one from 2016), CLS 5 (2007/2016), CLS 7 (2004/2019), CLS 8 (2004/2012), CLS 9 (2004/2016), CLS 13 (2004/2015), CLS 14 (2005/2013/2020), CLS 15 (2005/2008), CLS 19 (2005/2011), CLS 21 (2007/2018), CLS 24 (2005/2018), CLS 26 (2004/2018), CLS 28 (2005/2016), CLS 30 (2008/2015), CLS 47 (2004/2011/2019), CLS 49 (2006/2012/2015), CLS 50 (2005/2018), CLS 67 (2007/2013/2016), CLS 71 (2007/2020), CLS 91 (2005/2010), CLS 93 (2008/2019), CLS 119 (2012/2018), CLS 152 (2014/2020), CLS 157 (2011/2019), CLS 195 (2013/2017) and CLS 217 (2015/2017). Three of these clusters were independently studied by WGS: CLS 7⁷ (57 isolates sequenced), CLS 50⁹ (32 isolates sequenced), and CLS 217⁸ (13 isolates sequenced).

Thirteen genomes from different sub-lineages, six from L4.1.2 and seven from L4.3, not involved in any of the large outbreaks (non-clustered strains) and sequenced in previous studies, were used to make the comparisons with the clustered strains belonging to these sub-lineages.