Abstract—The gene expression reproducibility of microarray experiments has long been criticized. Many studies have been undergoing to address this issue. In this study we investigate the reproducibility problem when not only technical variance but also complex biological variances are involved by introducing the treatment of two different batches of Cordyceps sinensis (CS) to immature dendritic cells (DCs) which were isolated from blood samples of different individuals. Cordyceps sinensis, a complex compound, has been commonly used as herbal medicine and a health supplement in China for over two thousand years. In this study, we adopted duplicate sets of loop-design microarray experiments to examine two different batches of CS and analyze the effects of CS on DCs. Immature DCs were treated with CS, lipopolysaccharide (LPS), or LPS/CS for two days, and the gene expression profile were examined using microarrays. The results of two loop-design microarray experiments showed good intersection rates. The expression level of common genes found in both loop-design microarray experiments was consistent, and the R2 was higher than 0.93. Index Terms—Cordyceps sinensis, microarray, gene expression profile, reproducibility Manuscript received July 18, 2012; revised August 10, 2012. This work was supported in part by the Committee on Chinese Medicine and Pharmacy of the Department of Health (Taiwan) under Grant CCMP93-RD-051 and CCMP-94-RD-049. H.-T. Chang is with the Institute of Nanoengineering and Microsystem, National Tsing Hua University, Hsinchu 30013, Taiwan (e-mail: d929209@oz.nthu.edu.tw). C.-Y. Huang is with the Department of Biomedical Engineering and Environmental Sciences, National Tsing Hua University, Hsinchu 30013, Taiwan (e-mail: d924532@oz.nthu.edu.tw). C.-R. Chen is with the Department of Biomedical Engineering and Environmental Sciences, National Tsing Hua University, Hsinchu 30013, Taiwan (e-mail: d948504@oz.nthu.edu.tw). C.-W. Chang is with the Department of Biomedical Engineering and Environmental Sciences, National Tsing Hua University, Hsinchu 30013, Taiwan (e-mail: d924546@oz.nthu.edu.tw). W.-Y. Shu is with the Institute of Statistics, National Tsing Hua University, Hsinchu 30013, Taiwan (e-mail: shu@stat.nthu.edu.tw). C.-S. Chiang is with the Department of Biomedical Engineering and Environmental Sciences, National Tsing Hua University, Hsinchu 30013, Taiwan (e-mail: cschiang@mx.nthu.edu.tw). C.-Y. Li was with the Department of Biomedical Engineering and Environmental Sciences, National Tsing Hua University, Taiwan. He is now with the Department of Urology, University of Texas Southwestern Medical Center, Dallas, TX, 75390 USA (corresponding author phone: 214-516-8408; fax: 214-648-8786; e-mail: chiayangli@gmail.com). I. C. Hsu is with the Department of Biomedical Engineering and Environmental Sciences, National Tsing Hua University, Hsinchu 30013, Taiwan (corresponding author phone: +886-3-5727303; fax: +886-3-5710629; e-mail: ichsu@mx.nthu.edu.tw). I. INTRODUCTION ICROARRY technology is being applied widely to address increasingly complex scientific questions [1]. Microarray experiments yield lists of tens or hundreds of differentially regulated genes in sets of experiments. However, the presence of dissimilar regulatory patterns among functionally related genes makes it difficult for the biological interpretation of microarray data [2]. This is not surprising, because systematic biases and random variations are inherent in microarray data [1]. A careful experimental design and rigorous statistical analysis can increase the precision of microarray measurements [3], [4]. Moreover, statistical assessment is not only important in data analysis, but also plays a critical role in every stage of the microarray investigative process, including design of the experiment, data preprocessing, evaluation of systematic errors, identification of differentially expressed genes, functional classification, and biological interpretation [3], [4]. Kerr and Churchill first established the loop design for microarray experiments [5]. Previous studies demonstrated that loop design is more efficient than reference design because a range of statistical methods can be employed to increase the statistical power and robustness of microarray data analysis [6], [7]. Additionally, the loop-designed approach has a high hybridizations/nodes ratio that markedly increases the empirical power of microarray measurement [8]. The gene expression reproducibility of microarray experiments has long been criticized. In this study we investigate the reproducibility problem when not only technical variance but also complex biological variances are involved by introducing the treatment of Cordyceps sinensis (CS) to immature dendritic cells (DCs) isolated from blood samples of different individuals. Two kinds of replication are employed for the estimation of variance at different levels: technical and biological replicates. Technical replication is used to estimate system variance such as sample preparation and other effects of artifacts. Biological replication is used to evaluate variance in biological specimens and experimental procedures. Biological variance includes the heterogeneous distribution of cell types of both treating sample, in this study i.e. CS, and the treated sample, in this study i.e. immature DCs. Cordyceps sinensis (CS) is a species of parasitic fungus on Gene Expression Reproducibility Analysis for the Treatment of Two Different Batches of Cordyceps Sinensis by Loop-design Microarray Experiments Hung-Tsu Chang, Chao-Ying Huang, Chaang-Ray Chen, Cheng-Wei Chang, Wun-Yi Shu, Chi-Shiun Chiang, Chia-Yang Li, and Ian C. Hsu M Proceedings of the World Congress on Engineering and Computer Science 2012 Vol II WCECS 2012, October 24-26, 2012, San Francisco, USA ISBN: 978-988-19252-4-4 ISSN: 2078-0958 (Print); ISSN: 2078-0966 (Online) WCECS 2012