An effective feature extraction method is key to improving the accuracy of a prediction model. From the Gene Expression Omnibus (GEO) database, which includes 13,487 genes, we obtained microarray gene expression data for 238 samples from colorectal cancer (CRC) samples and normal samples. Twelve gene modules were obtained by weighted gene co-expression network analysis (WGCNA) on 173 samples. By calculating the Pearson correlation coefficient (PCC) between the characteristic genes of each module and colorectal cancer, we obtained a key module that was highly correlated with CRC. We screened hub genes from the key module by considering module membership, gene significance, and intramodular connectivity. We selected 10 hub genes as a type of feature for the classifier. We used the variational autoencoder (VAE) for 1159 genes with significantly different expressions and mapped the data into a 10-dimensional representation, as another type of feature for the cancer classifier. The two types of features were applied to the support vector machines (SVM) classifier for CRC. The accuracy was 0.9692 with an AUC of 0.9981. The result shows a high accuracy of the two-step feature extraction method, which includes obtaining hub genes by WGCNA and a 10-dimensional representation by variational autoencoder (VAE).
Background: Gastrointestinal cancer cells display both morphology and physiology diversity, thus posing a significant challenge for precise representation by a single data model. We conducted an in-depth study of gastrointestinal cancer heterogeneity by integrating and analyzing data from multiple modalities. Methods: We used a modified Canny algorithm to identify edges from tumor images, capturing intricate nonlinear interactions between pixels. These edge features were then combined with differentially expressed mRNA, miRNA, and immune cell data. Before data integration, we used the K-medoids algorithm to pre-cluster individual data types. The results of pre-clustering were used to construct the kernel matrix. Finally, we applied spectral clustering to the fusion matrix to identify different tumor subtypes. Furthermore, we identified hub genes linked to these subtypes and their biological roles through the application of Weighted Gene Co-expression Network Analysis (WGCNA) and Gene Ontology (GO) enrichment analysis. Results: Our investigation categorized patients into three distinct tumor subtypes and pinpointed hub genes associated with each. Genes MAGI2-AS3, MALAT1, and SPARC were identified as having a differential impact on the metastatic and invasive capabilities of cancer cells. Conclusion: By harnessing multimodal features, our study enhances the understanding of gastrointestinal tumor heterogeneity and identifies biomarkers for personalized medicine and targeted treatments.
The abstract should be a total of about 250 words and structured to contain the following headings: Background/Objectives, Methods, Results, Conclusions. Background/Objectives: Place the question addressed in a broad context and highlight the purpose of the study; Methods: Describe briefly the main methods or treatments applied. Include any relevant preregistration numbers, and species and strains of any animals used; Results: Summarize the article’s main findings; Conclusions: Indicate the main conclusions or interpretations. The abstract should be an objective representation of the article: it must not contain results which are not presented and substantiated in the main text and should not exaggerate the main conclusions. Clinical trial abstracts should include items that the CONSORT group has identified as essential.