| Hi-index | 0.00 |
Based on microarray data, a basic task is to extract differentially expressed (DE) genes between disease states and their associated functions to understand disease mechanisms. However, few such analyses have been conducted for breast cancer metastasis, possibly owing to the uncertainty of the disease state assignment for patients, which may lead to an extremely low power of detecting DE genes. In this study, we analyzed five datasets composed of metastatic and non-metastatic breast primary cancer samples. For two datasets in which few DE genes could be detected by the conventional false discovery rate control approach, a clustering approach was applied to select a group of genes with large differential expression changes between two groups of samples, in which the powers of identifying DE genes increased greatly. Then, we showed that each of the five DE gene lists captured a part of the differential expression signals from which we were able to extract metastasis-associated functions non-randomly reproducible across different datasets. Our results highlighted that many general biological processes (such as 'cell division', 'cell cycle', 'microtubule-based processes' and 'chromosome segregation'), rather than only their sub-processes, may be globally altered during the course of breast cancer metastasis, characterizing cancer metastasis as a 'systems disease'.