2.1. OSAnalyzer and SNPs Handling for Computing Kaplan-Meier
The identifiers of SNPs (i.e., A/G, G/A, and so on) typical for the OS-dataset represented in Table 2 are used to divide a generic probe into groups of samples that have the same SNP, but are not useful to estimate K and M. At most three groups of samples can be presented in each probe, given that the possible combinations of SNPs for each probe are equal to three (i.e., if the probe contains the alleles A and G then the possible combinations will be A/A, G/A, and G/G). In this way, for each probe it will be possible to group together the samples with the same polymorphism and evaluate the survival of the samples belonging to the identified groups. Before creating groups of samples identified by the same SNP, it is essential to link together all the SNPs and survival data for the same sample. This connection was realized by defining a virtual projection function (VPF) able to link together the survival data of each subject with their SNPs. The VPF virtually links together the survival data with all the SNPs belonging to the same sample without replicating the overall survival data for all the 1936 SNP for each subject, as conveyed in Figure 1.
At the same time this tight coupling between SNPs and the related survival data make it possible to split each probe into groups and accurately compute the K and M estimator for each identified group. By means of the VPF, OSAnalyzer is able to compute K and M by using only temporal data without needing to convert the literal SNPs to numerical values.