Table 2.

Signals of bacterial association in a pooled analysis

Gene ID	Gene name	Gene population frequency	Method	OR/effect size	p-value	Function
SPN23F05680	ldcB/dacB	0.81	Missense burden in invasive	1.20	1.3 × 10–19	D-alanyl-D-alanine carboxypeptidase (peptoglycan peptide precursors)
SPN23F22240	pspC/cbpA	0.59	K-mers	1.05	3.8 × 10–13	Binds secretory IgA, C3 and complement factor H; adhesin
SPN23F08080	spnTVR hsdS	1.00	Missense burden in invasive	1.08	2.8 × 10–6	Type I restriction-modification system specificity subunit S
SPN23F17820	psrP	0.42	Tajima’s D	−2.71 (invasive) −2.59 (carriage)	<1 × 10–6	adhesin
SPN23F10590	zmpD	0.53	Burden of rare LoF in invasive	1.42	<1 × 10–10	Unknown; paralogous to IgA1 protease (zmpA)
SPN23F09820	FM211187.3090	0.36	Missense burden in invasive	1.30	3.4 × 10–49	Bacteriocin precursor
SPN23F05670	FM211187.1804	1.00	Missense burden in invasive	1.27	4.3 × 10–47	Histidine triad family protein (nucleotide phosphatase)
SPN23F04740	ecsA	1.00	Missense burden in invasive	1.15	1.4 × 10–8	ABC transporter ATPase
SPN23F11460	mcrB	—	Missense burden in invasive	1.13	1.2 × 10–6	Endonuclease

Combining The Netherlands (meningitis only) and South African (all IPD cases) cohorts into a single dataset and performing a pGWAS with cohort as a covariate. Table shows genes significant (after applying a Bonferroni correction for the number of tests and association methods p < 0.05) in a pooled analysis of both cohorts with any of the association approaches, ordered by p-value. Odds ratios are with respect to carriage samples. The four genes in bold at top of the table are immunogenic and have previous evidence for association with virulence. For Tajima’s D, the effect size is the difference between D values, and for k-mers and LoF burden tests it is the odds ratio. For some p values, the calculation only allows an upper bound to be produced. The locus tag in the ATCC 700669 reference is listed, along with the common gene name if available

OR odds ratio