Dataset Information

Dissection of gene expression datasets into clinically relevant interaction signatures via high-dimensional correlation maximization.

ABSTRACT: Gene expression is controlled by many simultaneous interactions, frequently measured collectively in biology and medicine by high-throughput technologies. It is a highly challenging task to infer from these data the generating effects and cooperating genes. Here, we present an unsupervised hypothesis-generating learning concept termed signal dissection by correlation maximization (SDCM) that dissects large high-dimensional datasets into signatures. Each signature captures a particular signal pattern that was consistently observed for multiple genes and samples, likely caused by the same underlying interaction. A key difference to other methods is our flexible nonlinear signal superposition model, combined with a precise regression technique. Analyzing gene expression of diffuse large B-cell lymphoma, our method discovers previously unidentified signatures that reveal significant differences in patient survival. These signatures are more predictive than those from various methods used for comparison and robustly validate across technological platforms. This implies highly specific extraction of clinically relevant gene interactions.

SUBMITTER: Grau M

PROVIDER: S-EPMC6883077 | biostudies-literature | 2019 Nov

REPOSITORIES: biostudies-literature

ACCESS DATA

Publications

Dissection of gene expression datasets into clinically relevant interaction signatures via high-dimensional correlation maximization.

Grau Michael M Lenz Georg G Lenz Peter P

Nature communications 20191128 1

Gene expression is controlled by many simultaneous interactions, frequently measured collectively in biology and medicine by high-throughput technologies. It is a highly challenging task to infer from these data the generating effects and cooperating genes. Here, we present an unsupervised hypothesis-generating learning concept termed signal dissection by correlation maximization (SDCM) that dissects large high-dimensional datasets into signatures. Each signature captures a particular signal pat ...[more]

PMID: 31780653

Similar Datasets

Project description:Glioblastoma (GBM) heterogeneity, aggressiveness and infiltrative growth drastically limit success of current standard of care drugs and efficacy of various new therapeutic approaches. There is a need for new therapies and models reflecting the complex biology of these tumors to analyze the molecular mechanisms of tumor formation and resistance, as well as to identify new therapeutic targets. We established and screened a panel of 26 patient-derived subcutaneous (s.c.) xenograft (PDX) GBM models on immunodeficient mice, of which 15 were also established as orthotopic models. Sensitivity toward a drug panel, selected for their different modes of action, was determined. Best treatment responses were observed for standard of care temozolomide, irinotecan and bevacizumab. Matching orthotopic models frequently show reduced sensitivity, as the blood-brain barrier limits crossing of the drugs to the GBM. Molecular characterization of 23 PDX identified all of them as IDH-wt (R132) with frequent mutations in EGFR, TP53, FAT1, and within the PI3K/Akt/mTOR pathway. Their expression profiles resemble proposed molecular GBM subtypes mesenchymal, proneural and classical, with pronounced clustering for gene sets related to angiogenesis and MAPK signaling. Subsequent gene set enrichment analysis identified hallmark gene sets of hypoxia and mTORC1 signaling as enriched in temozolomide resistant PDX. In models sensitive for mTOR inhibitor everolimus, hypoxia-related gene sets reactive oxygen species pathway and angiogenesis were enriched. Our results highlight how our platform of s.c. GBM PDX can reflect the complex, heterogeneous biology of GBM. Combined with transcriptome analyses, it is a valuable tool in identification of molecular signatures correlating with monitored responses. Available matching orthotopic PDX models can be used to assess the impact of the tumor microenvironment and blood-brain barrier on efficacy. Our GBM PDX panel therefore represents a valuable platform for screening regarding molecular markers and pharmacologically active drugs, as well as optimizing delivery of active drugs to the tumor.

Dataset Information

Dissection of gene expression datasets into clinically relevant interaction signatures via high-dimensional correlation maximization.

Publications

Dissection of gene expression datasets into clinically relevant interaction signatures via high-dimensional correlation maximization.

Similar Datasets

OmicsDI is part of the ELIXIR infrastructure

Tweets