Partitioned twin analysis: a power study.

Individual differences in the human genome may now be measured with molecular genetic techniques. Therefore, dizygotic (DZ) twins may be classified as sharing two, one, or zero "genes" identical by descent for any measured polymorphism. As a result, we may partition genetic variation into two sources: (i) genotypes at and closely linked to particular marker loci identified with restriction fragment length polymorphisms (RFLPs) and (ii) other genetic variation. The power of the classical twin study to reject false models lacking either a marker effect or a residual genetic effect is explored. Additivity of genetic effects at or near the locus and of the residual genetic variation as well as random environmental variation are assumed. Results indicate that statistical rejection of models could be achieved with sample sizes which are within the range of several current twin registers. A design including monozygotic (MZ) twins is compared with one consisting of only DZ twins. MZ twins add considerable power for the detection of residual genetic variation but provide no information to resolve genetic marker effects.