1. ¹Faculty of Engineering, Tokyo University of Science, 1-3 Kagurazaka, Shinjyuku-ku, Tokyo 162-8601, Japan
2. ²Biostatistics and Data Management Department, Kowa Co., Ltd., Tokyo, Japan
3. ³Genetics Division, National Cancer Center Research Institute, Tokyo, Japan

Correspondence: Yasunori Sato, Faculty of Engineering, Tokyo University of Science, 1-3 Kagurazaka, Shinjyuku-ku, Tokyo 162-8601, Japan. Fax: +81-3-32605770. E-mail: yasu@ms.kagu.tus.ac.jp

Received 8 June 2004; Accepted 18 September 2004; Published online 24 November 2004.

Abstract

A genome-wide linkage equilibrium mapping is an emerging strategy to identify risk-modifying genes for common diseases, despite unsettled controversies upon many aspects, including its premises, designs, marker choices and cost benefits. One large-scale attempt in Japan aims to identify disease-associated single nucleotide polymorphisms (SNPs) for five diseases among the Japanese population: Alzheimer's disease, gastric cancer, diabetes, hypertension and asthma. Following an initial screening of c.a. 100,000 SNPs on 940 subjects (five diseases 188 patients) to select about 2,000 SNPs, we compared which subsequent screening design is more appropriate, and an additional one or two screens to further narrow down any disease-associated SNPs within a fixed total volume of 15,040,000 typings (2,000 SNPs five diseases 1,504 subjects, comprising 752 cases and 752 controls). We employed a Monte Carlo simulation to evaluate the probability of identifying truly disease-associated SNPs. The results suggest the single additional stage design (i.e., total two-stage design including the initial screening of 100,000 SNPs) was more practicable for the simple reason that the gain in probability is considered insufficient relative to an associated increase in study complexity in the three-stage design.