RESUMO
Machine Learning models have been frequently used in transcriptome analyses. Particularly, Representation Learning (RL), e.g., autoencoders, are effective in learning critical representations in noisy data. However, learned representations, e.g., the "latent variables" in an autoencoder, are difficult to interpret, not to mention prioritizing essential genes for functional follow-up. In contrast, in traditional analyses, one may identify important genes such as Differentially Expressed (DiffEx), Differentially Co-Expressed (DiffCoEx), and Hub genes. Intuitively, the complex gene-gene interactions may be beyond the capture of marginal effects (DiffEx) or correlations (DiffCoEx and Hub), indicating the need of powerful RL models. However, the lack of interpretability and individual target genes is an obstacle for RL's broad use in practice. To facilitate interpretable analysis and gene-identification using RL, we propose "Critical genes", defined as genes that contribute highly to learned representations (e.g., latent variables in an autoencoder). As a proof-of-concept, supported by eXplainable Artificial Intelligence (XAI), we implemented eXplainable Autoencoder for Critical genes (XA4C) that quantifies each gene's contribution to latent variables, based on which Critical genes are prioritized. Applying XA4C to gene expression data in six cancers showed that Critical genes capture essential pathways underlying cancers. Remarkably, Critical genes has little overlap with Hub or DiffEx genes, however, has a higher enrichment in a comprehensive disease gene database (DisGeNET) and a cancer-specific database (COSMIC), evidencing its potential to disclose massive unknown biology. As an example, we discovered five Critical genes sitting in the center of Lysine degradation (hsa00310) pathway, displaying distinct interaction patterns in tumor and normal tissues. In conclusion, XA4C facilitates explainable analysis using RL and Critical genes discovered by explainable RL empowers the study of complex interactions.
Assuntos
Inteligência Artificial , Neoplasias , Humanos , Genes Essenciais , Bases de Dados Factuais , Perfilação da Expressão GênicaRESUMO
Epithelial ovarian cancer (EOC) is the most common and fatal subtype of ovarian malignancies, with no effective therapeutics available. Our previous studies have demonstrated extraordinary suppressive efficacy of enterolactone (ENL) on EOC. A chemotherapeutic agent, trabectedin (Trabe), is shown to be effective on ovarian cancer, especially when combined with other therapeutics, such as pegylated liposomal doxorubicin or oxaliplatin. Thrombospondin 1 (THBS1), a kind of matrix glycoprotein, plays important roles against cancer development through inhibiting angiogenesis but whether it is involved in the suppression of EOC by ENL or Trabe remains unknown. To test combined suppressive effects of ENL and Trabe on EOC and possible involvement of THBS1 in the anticancer activities of ENL and Trabe. The EOC cell line ES-2 was transfected with overexpressed THBS1 by lentivirus vector. We employed tube formation assay to evaluate the anti-angiogenesis activity of ENL and of its combined use with Trabe after THBS1 overexpression and established drug intervention and xenograft nude mouse cancer models to assess the in vivo effects of the hypothesized synergistic suppression between the agents and the involvement of THBS1. Mouse fecal samples were collected for 16S rDNA sequencing and microbiota analysis. We detected strong inhibitory activities of ENL and Trabe against the proliferation and migration of cancer cells and observed synergistic effects between ENL and Trabe in suppressing EOC. ENL and Trabe, given either separately or in combination, could suppress the tube formation capability of human microvascular endothelial cells, and this inhibitory effect became even stronger with THBS1 overexpression. In the ENL plus Trabe combination group, the expression of tissue inhibitor of metalloproteinases 3 and cluster of differentiation 36 was both upregulated, whereas matrix metalloproteinase 9, vascular endothelial growth factor, and cluster of differentiation 47 were all decreased. With the overexpression of THBS1, the results became even more pronounced. In animal experiments, combined use of ENL and Trabe showed superior inhibitory effects to either single agent and significantly suppressed tumor growth, and the overexpression of THBS1 further enhanced the anti-cancer activities of the drug combination group. ENL and Trabe synergistically suppress EOC and THBS1 could remarkably facilitate the synergistic anticancer effects of ENL and Trabe.
Assuntos
Neoplasias Ovarianas , Trombospondina 1 , Animais , Camundongos , Humanos , Feminino , Carcinoma Epitelial do Ovário , Trabectedina/uso terapêutico , Trombospondina 1/uso terapêutico , Fator A de Crescimento do Endotélio Vascular , Células Endoteliais/metabolismo , Neoplasias Ovarianas/tratamento farmacológico , Neoplasias Ovarianas/patologia , Linhagem Celular Tumoral , Proliferação de Células/genéticaRESUMO
Approaches systematically characterizing interactions via transcriptomic data usually follow two systems: (i) coexpression network analyses focusing on correlations between genes and (ii) linear regressions (usually regularized) to select multiple genes jointly. Both suffer from the problem of stability: A slight change of parameterization or dataset could lead to marked alterations of outcomes. Here, we propose Stabilized COre gene and Pathway Election (SCOPE), a tool integrating bootstrapped least absolute shrinkage and selection operator and coexpression analysis, leading to robust outcomes insensitive to variations in data. By applying SCOPE to six cancer expression datasets (BRCA, COAD, KIRC, LUAD, PRAD, and THCA) in The Cancer Genome Atlas, we identified core genes capturing interaction effects in crucial pan-cancer pathways related to genome instability and DNA damage response. Moreover, we highlighted the pivotal role of CD63 as an oncogenic driver and a potential therapeutic target in kidney cancer. SCOPE enables stabilized investigations toward complex interactions using transcriptome data.