NEW YORK (GenomeWeb) – The European Commission has awarded an international group of companies and institutions nearly $11 million to develop a new platform for the predictive modeling of cancer.
Called CanPathPro, the consortium aims to combine omics data and systems biology tools into a single commercial platform for testing cancer hypotheses. Berlin-based Alacris Theranostics is coordinating the new five-year project, which will run through February 2021 and has been funded through the EC's Horizon 2020 research and innovation program.
"The aim is to have a commercial system in the end that can be used by other small- to medium-sized enterprises, biotechs, pharmaceutical companies, and academics, to make mechanistic predictions of their experimental systems," said Alacris CEO Bodo Lange.
To realize aims of the project, Alacris is working with partners at Mouse Clinical Institute in France; NKI, the Netherlands Cancer Institute; the Liebniz Institute on Aging in Germany; the German Research Center for Environmental Health; the Spanish National Research Council; and Simula Research Laboratory in Norway. The consortium also includes BiognoSys, a Swiss proteomics company based in Zurich; and Finovatis, a Lyon, France-based R&D company.
Lange told GenomeWeb that the partners formed CanPathPro to improve upon current approaches in cancer studies, which often rely on statistical and pattern recognition techniques for analyzing omics data, or, sometimes, modeling a single cellular signaling pathway.
Because of the way these approaches are set up, though, they typically fail to take into account the heterogeneity and complexity of cancer, limiting their utility for researchers, Lange said.
To address these limitations, CanPathPro intends to develop an entire suite of bioinformatics and experimental tools for the evaluation and control of systems biology modeling predictions. The resulting CanPathPro toolkit will include mouse and organotypic experimental systems, next-generation sequencing and quantitative proteomics data, plus a systems biology computational model for data integration, visualization, and predictive modeling.
Alacris will eventually offer these new tools via a combined experimental and systems biology platform that will allow its customers to integrate data sets to predict the activation status of individual pathways. That will in turn allow researchers to identify cancer signaling networks, as well as to come up with new hypotheses about biological systems, which can then be validated in experiments. The consortium members hope that the tools will help users predict cancer progression and drug efficacy, improving outcomes for their patients.
"Usually, what is done in cancer is statistical association," said Lange of existing approaches. "You look at previous drug response and correlate it with transcriptome or mutation profiles, and then you can stratify patients or mouse models in certain drug response," he said. "This is very limited because cancer is very complex, and there if you look for instance at even good biomarkers, only a subfraction of a population will have these biomarkers."
CanPathPro's vision, Lange said, is to develop a mechanistic model, that includes pathway information, protein-protein information, protein degradation, and protein phosphorylation data, all of it validated in high detail and tested on mouse models.
"The predictive modeling that has been done so far has focused on pathways or smaller systems, and here we are really looking at a very large system with thousands of components with detailed measurements to validate the system," said Lange.
The project "fits in nicely" with Alacris' focus as a company, he noted. The firm already offers a predictive modeling system called ModCell to partners, including GSK, that are focused on drug development and personalized medicine in oncology. ModCell relies on next-generation sequencing and other omics data, and exists within Alacris' IT infrastructure. As such, the toolkit generated by the CanPathPro project could be made available via Alacris' existing channels.
"We have a big interest to commercialize this," said Lange. "There are other companies [participating] that want to develop new technologies during the project and commercialize these, too," he added. Because of the nature of the project, each partner contributes and protects its own intellectual property, Lange said.
Each of CanPathPro's participants is focused on a different element of the project, and each is looking to benefit from the project in its own way.
NKI in the Netherlands and the Mouse Clinical Institute in France, for instance, are working together on expanding the information available on relevant mouse models. The consortium will examine the models using transcriptome exome sequencing, quantitative proteomics, and phosphoproteomics. "There will be very much information generated that hasn't been generated before on these mouse models," said Lange.
Yann Herault, director of the Strasbourg-based Mouse Clinical Institute, told GenomeWeb that by taking part in CanPathPro, his group aims to "modelize and study the complexity of the tumor growth and the organism response to therapeutics." He added that contributing to CanPathPro would position the Mouse Clinical Institute at the "forefront of cancer modeling and research."
Oliver Rinner, CEO of Schlieren, Switzerland-based BiognoSys, said that the firm will employ its next-generation proteomics platform to provide quantitative data points of protein expression and phosphorylation status that can later be inputted into Alacris' simulation infrastructure. Much of its contribution will be carried out using its Hyper Reaction Monitoring mass spectrometry platform, which it currently offers as a service. By using its platform in CanPathPro, Biognosys hopes to raise its profile, demonstrate its technology, and win over new clients.
"We believe that the ability to do real biology with the help of our data will drive the use and the demand for our services and products in the long run," Rinner told GenomeWeb.
The mass of transcriptome exome sequencing, quantitative proteomics, and phosphoproteomics data generated in the project will be the input for a number of participating bioinformatics teams. "They are looking at these data and integrating it with known data," said Lange. "The approach is to have a predictive model that can predict phenotypes and response data: a mechanistic model."
"The mathematical models used in this project can easily contain several thousands of coupled differential equations," Glenn Terje Lines, a senior research scientist at Lyksaker, Norway-based Simula Research Laboratory, and who is heading one of those bioinformatics teams, told GenomeWeb. "In the parameter estimation procedure, one has to solve these systems repeatedly, leading to a huge computational problem," he said. "We will develop algorithms and software that tackle these challenges, through the use of modern hardware such as many-integrated-core coprocessors, purpose-built solvers, and statistical tools."
While Alacris is clearly interested in making the resulting toolkit available to its biotech and pharma clients, as well as academics, Lange said the project should also especially benefit small- to medium-sized enterprises that do not have the resources to create their own comprehensive predictive models.
"In the end, a customer could come in, ask to have mice produced with certain mutations, and to have them analyzed to know how their systems are behaving," said Lange. "Customers could also come in with their own proteomics or transcriptome data and request predictions of how their systems will behave and what the phenotype will be if their favorite gene is modified or mutated. They will be able to save a lot of time because we can predict the outcome of their experiment."