From the point of accuracy improvement, our result is of concordance with the
results of other previous studies [37, 38]. It is interesting to compare the list GSK2126458 chemical structure of 15 genes selected by PAM and 8 genes as prior biological knowledge. In the current study, there was no overlap between these two gene lists, but the situation of overlap may be encountered in practice. Several genes may share the same or similar functions, so the existing of correlations among these genes from these two sources should be considered. Our result indicated that after the correlated gene had been added, no decrease of accuracy was found, which meant that there was no need to pay excess attention to the situation that overlapping existed between the information from microarray data and prior information. One of the main limitations for the present study
was how to incorporate prior biological knowledge and where to get it from. The prior biological knowledge in our study was retrieved from the literature, while, with the development of science and technology, huge knowledge will be discovered and reported. The magnitude of prior knowledge may have a certain impact on the results more or less. What information can be used as the truth and which kind of information should Selleckchem Vistusertib be excluded need to be further explored, maybe some experience could be borrowed from evidence-based medicine. On the other
hand, the minimum number of predictor genes is not known, which may serve as a potential limitation of the study, and the discrimination function can vary (for the same genes) based on the location and protocol used for sample preparation . The complexity of discriminant analysis and the multiple choices among the available discriminant methods are quite difficult tasks, which may influence the adoption by the clinicians in the future. Although highly accurate, microarray data’s widespread clinical relevance and applicability are still unresolved. Conclusion In summary, a simple and general framework to incorporate prior knowledge into discriminant analysis was proposed. Our method seems to be useful for Leukocyte receptor tyrosine kinase the improvement of classification accuracy. This idea may have good future not only in practice but also in methodology. Acknowledgements This study was partially see more supported by Provincial Education Department of Liaoning (No.2008S232), Natural Science Foundation of Liaoning province (No.20072103) and China Medical Board (No.00726.). The authors are most grateful to the contributors of the dataset and R statistical software. Peng Guan was supported by the Scientific Research Foundation for the Returned Overseas Chinese Scholars, State Education Ministry (No. 890) and a CMU Development grant (No. 5). References 1.