Unsupervised Feature Selection for Tumor Profiles using Autoencoders and Kernel Methods
This work addresses the challenge of discovering tumor subtypes without labeled data in cancer research, though it is incremental as it builds on existing unsupervised feature selection techniques.
The authors tackled the problem of high-dimensional, noisy tumor gene expression data by proposing an unsupervised feature selection method called LKFS, which combines autoencoders and kernel methods to select a subset of genes and improve clustering of tumor subtypes, resulting in lower redundancy and better clustering performance compared to benchmarks on brain, renal, and lung tumor datasets.
Molecular data from tumor profiles is high dimensional. Tumor profiles can be characterized by tens of thousands of gene expression features. Due to the size of the gene expression feature set machine learning methods are exposed to noisy variables and complexity. Tumor types present heterogeneity and can be subdivided in tumor subtypes. In many cases tumor data does not include tumor subtype labeling thus unsupervised learning methods are necessary for tumor subtype discovery. This work aims to learn meaningful and low dimensional representations of tumor samples and find tumor subtype clusters while keeping biological signatures without using tumor labels. The proposed method named Latent Kernel Feature Selection (LKFS) is an unsupervised approach for gene selection in tumor gene expression profiles. By using Autoencoders a low dimensional and denoised latent space is learned as a target representation to guide a Multiple Kernel Learning model that selects a subset of genes. By using the selected genes a clustering method is used to group samples. In order to evaluate the performance of the proposed unsupervised feature selection method the obtained features and clusters are analyzed by clinical significance. The proposed method has been applied on three tumor datasets which are Brain, Renal and Lung, each one composed by two tumor subtypes. When compared with benchmark unsupervised feature selection methods the results obtained by the proposed method reveal lower redundancy in the selected features and a better clustering performance.