CORNETO: extracting meaningful insights from your complex omics data
A new computational tool helps researchers map molecular interactions, revealing the pathways associated with disease.
Researchers at the European Molecular Biology Laboratory’s European Bioinformatics Institute (EMBL-EBI; Hinxton, UK) and Heidelberg University (Germany) have collaborated to develop CORNETO, a computational tool that uses machine learning to extract key molecular information from complex omics data. By extracting molecular network information, researchers can better understand the mechanisms underlying healthy and diseased cells.
Understanding how molecules interact inside our cells is key to uncovering how disrupted molecular pathways lead to disease. To assess molecular interactions, omics techniques are often used. Omics investigations, which encompass transcriptomics, genomics, proteomics and metabolomics, provide an immense amount of information about the molecular networks within living systems via a suite of high-throughput technologies. However, with advancements in omics technology leading to an increase in the complexity and size of omics data available, gleaning meaningful insights from the data can often prove challenging.
To overcome the analysis challenges associated with omics data, the research team developed CORNETO, which stands for Constrained Optimization for the Recovery of NETworks from Omics. This adaptable computational tool combines machine learning techniques with biological prior knowledge to simultaneously analyze multiple types of omics data, allowing researchers to map how genes, proteins and signaling pathways interact.
“We wanted to solve a common challenge in systems biology: how to make sense of omics data when you have so much complex data available all at once,” commented senior author Julio Saez-Rodriguez, Head of Research at EMBL-EBI and Professor on leave at Heidelberg University. “CORNETO helps by combining these complex data with prior information coming from biological databases to find patterns that are consistent, interpretable, and biologically meaningful.”
Mathematical modeling reveals the rules governing tissue organization
Every day, your body replaces billions of cells – and yet, your tissues stay perfectly organized. How is that possible?
Furthermore, by analyzing multiple samples or conditions simultaneously, CORNETO can highlight both shared and disparate biological processes across datasets, contributing to a much bigger and fuller picture of what’s happening in these samples.
To demonstrate the potential applications of this platform, the team deployed CORNETO in cancer research. They first tested CORNETO’s performance against synthetic data using different ‘regularization’ settings. Regularization in this context refers to the degree of similarity that the software selects for in the inferred signaling networks; increased regularization encourages the identification of common signaling pathways, with the trade-off being that each patient’s network would be less uniquely fitted to their individual dataset.
The researchers then used CORNETO to analyze gene expression data from several patients with lung adenocarcinoma to determine abnormalities in intracellular signaling pathways, comparing single- and multi-sample analyses. At lower regularization values, the single-sample approach yielded more total network connections per sample; however, at varying degrees of regularization, the multi-sample approach identified a higher proportion of interactions shared among patients and driven by the same group of deregulated kinases – enzymes that regulate cell signaling.
In addition to CORNETO’s use in lung adenocarcinoma research, the computational tool has been used to investigate yeast survival and growth as well as deregulated pathways associated with chemotherapy resistance in ovarian cancer patients.
CORNETO is open-source software, available on GitHub. For more information, including tutorials and example datasets, click here.
