Identification and Validation of the SNV Biomarkers Based on Multi-Dimensional Patterns.

Background: Single nucleotide variants (SNVs) are detected as different distributions of DNA samples of distinct types of cancer patients. Even though, it is an exacting task to select the appropriate method to identify cancer to the greatest extent of SNVs. Results: In this paper, we proposed a biomarker concept based on SNV patterns in different feature dimensions. Raw dataset (2761 samples) consisting of twelve different cancers was obtained from TCGA (The Cancer Genome Atlas). After preliminary screening of 562,321 DNA mutation sites in the samples, the mutation sites were extracted and characterized by cancer types in six different SNV feature dimensions. In this study, we found that the extracted features showed similar distribution in the cluster center of the disease type of the samples. After the initial processing of the raw data, the sample was more focused on the subtype distribution of the cancer or the cancer at the SNV level. We used k-nearest neighbors (KNN) to classify the extracted features and Leave-One-Out cross verified them. The accuracy of classifying is stable at around 97% and reached 97.43% at the highest. During the validation phase, we found validated oncogenes in the loci of the features with the highest importance among nine cancers. Conclusions: In summary, the samples showed consistent patterns according to the cancer in which it belongs. It is feasible to classify the cancer of the sample by the distribution of different dimensions of the SNVs and has a high accuracy. And has potential implications for the discovery of cancer-causing genes.