Table 2.

The average GDT-TS scores of top 1 models or best of five models predicted for 91 domains by MULTICOM_refine and NBIS-AF2-standard.

	Top 1 Model			Best of Five Models
Domain Type	MULTICOM_refine	NBIS-AF2-standard	p-value	MULTICOM_refine	NBIS-AF2-standard	p-value
All 91 domains	0.7922	0.7525	0.0006692	0.8185	0.7736	5.156e-06
46 FM & FM/TBM domains	0.6939	0.6449	0.01347	0.7335	0.6746	0.0003845
45 TBM domains	0.8927	0.8625	0.009202	0.9054	0.8749	0.001516

The p-value for the difference between MULTICOM_refine and NBIS-AF2-standard is calculated with the one-sided Wilcoxon signed rank test. In all the situations, MULTICOM_refine performs significantly better than NBIS-AF2-standard.