Scalable summary-statistics-based heritability estimation method with individual genotype level accuracy

Table 2.

Calibration of the methods

Method	Sample	BOLT-REML	GCTA-GREML	LDSC	SumHer-GCTA	SUM-RHE	RHE
Bias	10k	0.0247	0.0219	0.0084	0.0095	0.0058	0.0047
Bias	50k	—	—	0.0019	0.0005	−0.0002	−0.0002
SE	10k	0.0020	0.0020	0.0101	0.0031	0.0045	0.0045
SE	50k	—	—	0.0019	0.0006	0.0010	0.0010
FPR	10k	0.15	0.14	0.04	0.06	0.06	0.05
FPR	50k	—	—	0.03	0.05	0.06	0.06

We report the bias, SE, and the false-positive rate (FPR) of each method in the setting in which $\text{[math]}$ . Because of computational limitations, GCTA-GREML and BOLT-REML were only run on 10k samples.

Published in Advance July 22, 2024, doi: 10.1101/gr.279207.124 Genome Res. 2024. 34: 1286-1293

Scalable summary-statistics-based heritability estimation method with individual genotype level accuracy