Skip to main content
. 2018 Feb 26;10(3):266. doi: 10.3390/nu10030266

Figure 1.

Figure 1

(A) Workflow for designing genome-based personalized health care program. Among the set of SNPs related to obesity under GWAS catalog, SNPs of which has significant interaction effects for diet or exercise were assigned to four different effectiveness categories using generalized linear model: effectiveness of “carbohydrate intake changes” (CE), effectiveness of “fat intake changes” (FE), effectiveness of “total calories intake changes” (TE) or “effectiveness of exercise” (EE). Genetic risk scores (GRS) of four effectiveness categories were calculated per individual. The distributions of GRSs for each of four effectiveness classes were generated using the KoGES cohort. The degree of effectiveness was defined as “very low” (VL; smaller than the 25th percentile GRS), “low” (L; larger than or equal to the 25th percentile and smaller than the 50th percentile GRS), “high” (H; larger than or equal to the 50th percentile and smaller than the 75th percentile GRS), and “very high” (VH; larger than or equal to the 75th percentile GRS) for all of four GRS distributions; (B) Calculation of genetic risk score (GRS). The GRS of an individual is calculated based on the genotypes of the selected SNPs (column names of the table). The score of each SNP (3rd row in the table) was calculated by multiplying the genotype that is the number of risk alleles of each SNPs (1st row in the table) by the effect sign (2nd row in the table) which is determined by the sign of estimated coefficient of the corresponding interaction term from the linear model. These scores are summed over all the selected SNPs to generate the final effectiveness score of the individual (denoted as “Total score” at the right corner of the table). This process was applied to all of the individuals, as well as all of the four categories: “effectiveness of carbohydrate intake changes” (CE), “effectiveness of fat intake changes” (FE), “effectiveness of total calories intake changes” (TE), and “effectiveness of exercise” (EE). Thus, a total of four GRS distributions were generated.