Extended Data Fig. 1: Flowchart of sputum collection, quality control and microbiome analyses.

A total of 3,915 individuals were initially approached. 3,820 individuals completed the questionnaire. Of them, 3,737 individuals had body measurement (height and weight), and 3,424 individuals further had acceptable spirometry measurement after quality control (grade C or above). 907 individuals with post-bronchodilator FEV1 below 35% predicted, unwilling or unable to participate, or unable to produce sputum after induction were excluded, leading to 2,517 sputum samples. Of them, 84 individuals who reported the antibiotic use within one month and 782 individuals who yielded low quality (<0.2 g) sputum or yielded sputum that did not pass quality control were excluded, leading to 1,651 sputum samples stored and transported for further microbiome analysis. Genomic DNA was extracted from the 1,651 samples for 16S rRNA gene and metagenomic sequencing. For 719 samples with remaining genomic DNA and yielded ITS amplicon of adequate quality, ITS sequencing was further performed. For 1,651 sputum samples with metagenomic data, 1,128 had microbial coverage >70% as estimated by Nonpareil software, and were retained for further analyses.