Table 2.

Effect of mutations in hcn4 on sinoatrial pauses in data from the CRISPR/Cas9 F₁ and sa11188 experiments combined.

Age	Model	Study	nunaffected	naffected	OR	LCI	UCI	P
2dpf	Additive	CRISPR/Cas9 F1	258	39	2.749	1.461	5.171	1.7 × 10–3
		sa11188	394	12	1.697	0.695	4.148	2.5 × 10–1
		Combined	678	51	2.245	1.358	3.710	1.6 × 10–3
	2 vs. 0	CRISPR/Cas9 F1	127	23	3.667	0.633	21.249	1.5 × 10–1
		sa11188	193	7	2.529	0.468	13.657	2.8 × 10–1
		Combined	414	30	2.916	0.851	9.993	8.9 × 10–2
5dpf	Additive	CRISPR/Cas9 F1	312	9	2.460	0.680	8.899	1.7 × 10–1
		sa11188	308	34	2.252	1.303	3.893	3.7 × 10–3
		Combined	620	43	2.762	1.320	5.782	7.0 × 10–3
	2 vs. 0	CRISPR/Cas9 F1	231	4	–	–	–	–
		sa11188	151	16	7.885	1.719	36.163	7.9 × 10–3
		Combined	382	20	9.015	3.348	24.277	1.4 × 10–5
Any	Additive	Combined	631	90	2.121	1.456	3.090	8.9 × 10–5
	2 vs. 0	Combined	391	47	3.687	1.651	8.232	1.5 × 10–3

Associations between the presence of sinoatrial pauses during the 30 s recording and the number of mutated alleles in hcn4 weighted by the predicted effect of those mutations on protein function (additive); or in embryos with nonsense mutations in both hcn4 alleles vs. embryos free from CRISPR/Cas9-induced or sa11188 mutations in hcn4 (2 vs. 0) at 2 days post-fertilization (dpf), 5dpf, or at either time point (any). In the CRISPR/Cas9 experiment, associations were examined using logistic regression, adjusting for time of day, batch and for the weighted number of mutated alleles in the other genes as fixed factors; in the sa11188 experiment and combined analysis, associations were examined using mixed models (xtmelogit in Stata), with embryos nested in batches and experiments and adjusting for time of day as a fixed factor. At 2dpf, 26 and 94 unaffected larvae were dropped from the additive and ‘2 vs. 0’ analyses in the CRISPR/Cas9 F1 larvae due to multicollinearity. These observations were included in the combined analysis.