Table 1.

Comparison of TWAS scan discoveries of height GWAS from GIANT Consortium using GTEx whole blood data

Methods	PTWAS	Fusion	PrediXcan	SMR
Discoveries	1864 (490)	1813 (451)	1764 (437)	1693 (416)

We construct eQTL weights for PTWAS, TWAS-Fusion, PrediXcan, and SMR and perform TWAS scan using the summary statistics of the height GWAS data from the GIANT consortium. For all methods, the analysis focuses on 12,041 autosomal eGenes identified from the whole blood data. The association testing is carried out using the GAMBIT. The table shows the numbers of gene discoveries at the 5% FDR level (using the R package qvalue) by various methods. The numbers in the parentheses indicate the rejections after a much more stringent Bonferroni correction. In both cases, methods accounting for allelic heterogeneity empirically outperform the single-SNP-based SMR approach, and the proposed PTWAS scan yields the highest number of discoveries. This observation is markedly consistent with what is observed in our simulation studies