Table 3.
Heritability and Genetic Correlation Based on Different Datasets
| Method | #SNPs | Data |
# Individuals |
h2BMI |
h2SCZ (Liability Scale) |
Genetic Correlation |
|||||
|---|---|---|---|---|---|---|---|---|---|---|---|
| Mean | SD | Estimate | SE | Estimate | SE | Estimate | SE | p | |||
| GREML | 518,992 | UKBB+SCZ(qced) | 152,961 | – | 0.184 | 3.80E−03 | 0.192 | 4.39E−03 | −0.136 | 1.74E−02 | 4.54E−15 |
| LDSC | 516,519 | UKBB+SCZ(qced) | 151,262 | 1,432.7 | 0.255 | 1.38E−02 | 0.280 | 1.63E−02 | −0.173 | 3.08E−02 | 1.91E−08 |
| LDSC-meta1 | 477,163 | UKBB+GIANT+PGCSCZ | 422,499 | 20,226.0 | 0.111 | 8.10E−03 | 0.259 | 1.28E−02 | −0.091 | 2.44E−02 | 1.95E−04 |
| LDSC-meta2 | 1,011,748 | UKBB+GIANT+PGCSCZ | 414,707 | 32,697.8 | 0.121 | 6.50E−03 | 0.261 | 1.03E−02 | −0.087 | 1.90E−02 | 4.91E−06 |
GREML: Analysis was based on quality controlled genetic data for BMI (from UK Biobank with 111,019 individuals and 518,992 SNPs) and schizophrenia (from PGC with 41,630 individuals and 518,992 SNPs).
LDSC: The datasets used in LDSC were the same as in GREML.
LDSC-meta1: GWAS summary statistics for BMI were based on meta-analyzed GWAS results of UKBB individual-level genetic data (with 111,019 individuals and 518,992 SNPs) and of GIANT (245,051 individuals and 477,163 SNPs). For SCZ, the GWAS summary statistics from the full PGC sample based on 77,096 individuals were used.
LDSC-meta2: The datasets used in LDSC-meta2 were the same as in LDSC-meta1 except the increased number of SNPs (1,011,748) with which its performance was to check.
Mean and SD of #individuals: Due to different call rates of each SNP, number of individuals for each SNP used in GWAS were different.