Population Stratification Analysis in Genome-Wide Association Studies - Mathematical Approaches to Polymer Sequence Analysis and Related Problems - page 185

a

b

Case
Control

Case
Control

−0.05 0.00 0.05 0.10 0.15 0.20 −0.05 −0.03 −0.01 0.01

Eigenvector 1 Eigenvector 1

c

d

African
Asian
Caucasian

Africa
Europe
Other

−0.05 0.00 0.05 0.10 0.15 0.20 −0.05 −0.03 −0.01 0.01

Eigenvector 1 Eigenvector 1

Fig. 9.1 The top two principal components (eigenvectors 1 and 2) for 107 patients with
schizophrenia and 91 healthy controls of UCI sample ( a , c ) and for 741 cases and 741 controls
of CATIE-NIMH sample ( b , d )

the Europeans, green the Africans and blue the other or more than ethnicities in the
CATIE-NIMH sample. A significant p-value makes us reject the null hypothesis of
similarity between (a) cases and controls (CA/CO) and (b) ethnic groups, in both
UCI and CATIE-NIMH samples.
On the other hand, if we consider the samples according to ethnicities, ANOVA
statistics points to three major axis of variation in both samples (Table 9.2 b). The top
two axes of variation (PC1 and PC2), representing most of the genetic variance in the
UCI sample (p<10 12 ), are shown in Fig. 9.1 c. Interestingly, individuals clustered
in accordance with their ethnic origins: in particular, PC1 provided good separation
between African-Americans and Caucasians, while PC2 separated Caucasians and
Asians. In the CATIE-NIMH sample, as shown in Fig. 9.1 c, there is a subdivision
along the PC1 between European and Africans.
We represent the scree plot of the eigenvalues of PCA to evaluate which are the
PCs that describe the largest genetic variance and to confirm the ANOVA results.