Three Important Studies for Cancer Gene Diagnosis

This Chapter introduces three studies to confirm the correctness of cancer patients’ design three principles using four different sizes of microarrays. The first study (Study-1) evaluates three different sizes of microarrays, such as Liver3, Breast6, and Colorectal6. Colorectal6, having 63 patients, a medium-size array, has many “vital BGSs with M2 = 0 and less than five genes.” Study-2 analyzes Colorectal5 because it is almost the same size as Colorectal6. However, M2’s result of Colorectal5 is bad. Moreover, there are four misclassifications on the PCA scatterplot (Validation3). I omitted these four patients as test samples and got a great result from the modified Colorectal5 (44 patients (22:22)). Study-3 omitted the misclassifications by Validation3 and made five new microarrays, Liver300 (150:150), Liver200 (100:100), Liver100 (50:50 cancers), Liver60 (30:30), and Liver40 (20:20). Only Liver60 and Liver40 have many vital BGSs. These three studies conclude that the medium sample size microarrays are proper for cancer diagnosis because we can find many vital BGSs from those microarrays. Physicians can characterize vital BGSs by the legacy oncogenes included in them. Therefore, I propose new principles for cancer patients limited to gene diagnosis. These results open a new frontier of multivariate oncogenes for cancer gene diagnoses.