Dataset Information

Network modeling of patients' biomolecular profiles for clinical phenotype/outcome prediction.

ABSTRACT: Methods for phenotype and outcome prediction are largely based on inductive supervised models that use selected biomarkers to make predictions, without explicitly considering the functional relationships between individuals. We introduce a novel network-based approach named Patient-Net (P-Net) in which biomolecular profiles of patients are modeled in a graph-structured space that represents gene expression relationships between patients. Then a kernel-based semi-supervised transductive algorithm is applied to the graph to explore the overall topology of the graph and to predict the phenotype/clinical outcome of patients. Experimental tests involving several publicly available datasets of patients afflicted with pancreatic, breast, colon and colorectal cancer show that our proposed method is competitive with state-of-the-art supervised and semi-supervised predictive systems. Importantly, P-Net also provides interpretable models that can be easily visualized to gain clues about the relationships between patients, and to formulate hypotheses about their stratification.

SUBMITTER: Gliozzo J

PROVIDER: S-EPMC7046773 | biostudies-literature | 2020 Feb

REPOSITORIES: biostudies-literature

ACCESS DATA

Publications

Network modeling of patients' biomolecular profiles for clinical phenotype/outcome prediction.

Gliozzo Jessica J Perlasca Paolo P Mesiti Marco M Casiraghi Elena E Vallacchi Viviana V Vergani Elisabetta E Frasca Marco M Grossi Giuliano G Petrini Alessandro A Re Matteo M Paccanaro Alberto A Valentini Giorgio G

Scientific reports 20200227 1

Methods for phenotype and outcome prediction are largely based on inductive supervised models that use selected biomarkers to make predictions, without explicitly considering the functional relationships between individuals. We introduce a novel network-based approach named Patient-Net (P-Net) in which biomolecular profiles of patients are modeled in a graph-structured space that represents gene expression relationships between patients. Then a kernel-based semi-supervised transductive algorithm ...[more]

PMID: 32107391

Similar Datasets

Project description:Background Published multi-gene classifiers suggested outcome prediction for patients with stage UICC II colon cancer based on different gene expression signatures. However, there is currently no translation of these classifiers for application in routine diagnostic. Therefore, we aimed at validating own and published gene expression signatures employing methods which enable RNA and protein detection in routine diagnostic specimens. Results Immunohistochemistry was applied to 68 stage UICC II colon cancers to determine the protein expression of five selected previously published classifier genes (CDH17, LAT, CA2, EMR3, and TNFRSF11A). Correlation of protein expression data with clinical outcome within a 5-year post-surgery course failed to separate patients with a disease-free follow-up [Group DF] and relapse [Group R]). In addition, RNA from macrodissected tumor samples from 53 of these 68 patients was profiled on Affymetrix GeneChips (HG-U133 Plus 2.0). Prognostic signatures were generated by Nearest Shrunken Centroids with cross-validation. Although gene expression profiling allowed the identification of differentially expressed genes between the groups DF and R, a stable classification and prognosis signature was not discernable in our data. Furthermore, the application of previously published gene signatures consisting of 22 and 19 genes, respectively, to our gene expression data set using ‘global tests’ and leave-one-out cross-validation was unable to predict clinical outcome (prediction rate 75.5% and 64.2%; n.s.). T-stage was the only independent prognostic factor for relapse in multivariate analysis with established clinical and pathological parameters including microsatellite status. Conclusions Our protein and gene expression analyses currently do not support application of molecular classifiers for prediction of clinical outcome in routine diagnostic as a basis for patient-orientated therapy in stage UICC II colon cancer. Further studies are needed to develop prognosis signatures applicable in patient care.

Dataset Information

Network modeling of patients' biomolecular profiles for clinical phenotype/outcome prediction.

Publications

Network modeling of patients' biomolecular profiles for clinical phenotype/outcome prediction.

Similar Datasets

OmicsDI is part of the ELIXIR infrastructure

Tweets