CM-Predict: A Classification Model to Predict Cancer Metastasis Based on Co-expression Network

Cancer metastasis is a complex process that usually shows a preference for certain organs, with the brain, bone, liver, and lung being the four most common metastatic organs. Studying the metastatic characteristics of cancer with great importance for the clinical diagnosis of cancer and the treatment of patients with advanced cancer. In this paper, we propose CM-Predict, a method for predicting metastatic organs in cancer. Take advantage of the feature that the co-expression network of genes with the same type of cancer is consistent, build a framework for feature extraction using reference networks and perturbation networks, and use the extracted features to classify samples with cancer metastasis. We compared CM-Predict with other classification models, for the cancer metastasis classification task, CM-Predict significantly outperformed the four machine learning methods in the other models in the BLCA, ESCA, and LIHC datasets. We also used a new statistical analysis method for enrichment analysis of the features screened by CM-Predict, which provides a new means to elucidate the mechanisms of cancer metastasis.