What do we do?
Personalised medicine is centred on the general objective to integrate a patient’s own data with knowledge about basic genetic and biological mechanisms of disease to identify her own disease activation pathways and based on them, potentially
new therapeutic strategies uniquely suited to that patient. We build mathematical models to integrate knowledge on patient genetic abnormalities, on tumor drivers, and pharmacokinetics of drug combinations, to gain insights into the
vulnerabilities of the tumor signaling network.
Who we are
We are a group of computational scientists, modellers, and statisticians looking to apply network science methods to construct and analyse comprehensive disease- and patient-specific interaction networks. Our methods are in network analytics,
data science, and bioinformatics.
Objectives
Our objective is in developing methods to identify optimal drug therapy solutions customised on the patient’s own data. We integrate the genetic
abnormalities of a patient with the key disease drivers to identify efficient drug combinations for her case. Our current studies include glioblastoma (the most common and lethal form of central nervous system cancer) and multiple myeloma
(cancer of the plasma cell and the second most common blood cancer).
ABOUT US
Combining network science with big data analytics
Cancer is a dynamic, complex, multiscale disease with a major impact on society. Current estimates are that 38.4% of people may be diagnosed with it at some point during their lifetimes. It is a dynamic complex system following rules
that are in part known and in part can be deduced based on patient tumor- and treatment-data. This project is driven by the personalization of medicine through mathematics, modeling, and simulation. Mathematical models can integrate
knowledge on patient genetic abnormalities, on tumor drivers, and pharmacokinetics of drug combinations, to build comprehensive disease- and tumor-specific models. Such models can then be analyzed as dynamic complex systems based on
directed graphs to gain insights into the vulnerabilities of the tumor signaling network.
We address in this project two main research questions:
— Can patient- and disease-specific genetic data predict optimal drug combinations in oncology?
— Can molecular changes throughout an oncologic treatment inform the early detection of drug resistance and predict optimal treatment changes to avoid it?
This project is supported by a grant of the Romanian Ministry of Education and Research, CCCDI – UEFISCDI.
OUR TEAM
Project Director
Dr. Ion Petre (ion.petre@gmail.com)
Senior Researchers
Dr. Eugen Czeizler
Dr. Andrei Paun
Dr. Mihaela Paun
Dr. Victor Mitrana
Researchers
Dr. Geffry Barad
Daniela Florea
Dr. Georgiana Gavril
Dr. Ana-Maria Gheorghe
Dr. Corina Itcus
Dr. Octavian Pacioglu
Dr. Manuela Sidoroff
Dr. Romica Trandafir
Dr. Iris Tusa
PhD Students
Nicoleta Siminea
Victor Popescu
Laura Popa
David Pacioianu
KEY EXPERTISE
1.
Network science
We are experts in network analytics, especially network controllability, topological analysis, Boolean network dynamics.
2.
Large-scale disease-specific interaction networks
We build and analyse comprehensive, multi-source disease- and patient-specific networks, consisting of thousands of genes and the interaction between them.
3.
Network controllability
Our platform NetControl4BioMed can be used for the network controllability analysis of large-scale networks, with a focus on discovering efficient drug combinations in disease networks.
Objectives
This project focuses on the problem of controlling a dynamic patient-specific disease network. The basic problem setup we work on is that of a network where control is sought over a given set of targets, in the sense of being able to change their configuration through external interventions on some well-chosen input nodes in the network, taking advantage of the network topology. We are interested in finding a minimal set of input nodes in the network such that the behavior of the target nodes may be changed arbitrarily through a well-chosen sequence of signals to the input nodes, cascaded throughout the network through its wiring. We focus on formalizations of this network controllability problem that maximize its applicability in biomedicine, including a specific set of targets to choose from (e.g., disease-specific essential genes), a specific set of inputs to choose from (e.g., drug targets), as well as non-linear network topologies. This is a highly relevant research direction in synergistic network pharmacology and systems medicine, with drug repurposing and finding novel drug combinations central research objectives. Our proposal is on computational methods for solving optimization problems motivated and driven by these challenges.
Objective 1. To develop efficient algorithmic solutions for clinically-constrained models of linear network controllability.
We will give the network controllability problem realistic formulations, considering some of the constraints stemming from the applications in the biomedical domain. We will include the influence of off-target drug effects and measures of synergy for drug combinations. This will allow us to formulate questions related to finding optimal partners for a given drug. We will also formulate and investigate network controllability as a multi-objective optimisation problem, with the goal of applying it in the study of simultaneous targeting of tumour sub-clones, each with its own, potentially different molecular interaction network.
OBJECTIVE O2. TO EXPLORE TWO CANCER CASE-STUDIES TO INFORM THE WORK ON OBJECTIVE O1.
Glioblastoma is the most common and lethal form of central nervous system cancer. Despite continued efforts over decades to develop new treatments, none has appreciably improved how long patients live: most people with this type of brain cancer survive only about 15 months. Multiple myeloma is cancer of the plasma cell and the second most common blood cancer. Without treatment, typical survival is 7 months, while with current treatments it goes to around 4-5 years, with often difficult side-effects and high costs. We focus in this project on demonstrating the potential of network controllability for personalised medicine in these two types of cancer. On both case studies we have already identified rich data sets to support our research, as well as relevant collaborators.
Project stages
The project advances in 3 stages (08-12/2020, 01-12/2021, 01-07/2022), each with its own activities.
Stage 1: Data collection
08-12/2020
The objective for this stage of the project was to collect 100 patient genomic datasets for each of the two case studies (multiple myeloma and glioma) and data on drug synergies for the standard therapy drugs on multiple myeloma and glioma.
All of the data was anonymised in previous studies and was freely available online.
The glioma data came from the TCGA database. We selected RNA data to offer information on the tumour expression level. We identified data on 162 patients with glioma and 5 samples of healthy tissue. For each of the patients we also got
their treatment data: drug therapies, surgeries, chemo- and radiation-therapy, treatment response.
The multiple myeloma data came from the study by Lohr et al (Widespread genetic heterogeneity in multiple myeloma: implications for targeted therapy. Cancer Cell, 2014). The data is on 203 patients from USA and UK. For each of them, the
whole multiple myeloma genome was sequenced and its specific mutations identified.
For the drug synergy data we identified and analysed 6 different resources. Overall, they offer experimental data, literature data, user-uploaded data, computational AI-based data. All of them are offered as web-based services, some of
them as free and open resources.
Stage 2: Network synthesis
01-12/2021
The objectives of this stage of the project were to synthesise one interaction network for each of the patient in eacf of the case studies and to have the NetControl4BioMed software extended towards including synergy data and multiple control
strategies.
For each of the patient dataset in our project we generated a directed protein-protein interaction network, focused on the interactions around its own specific genetic abnormalities, including mutations and abnormal expression levels. Additionally,
we also included disease-specific nodes of interest, including typically mutated genes, disease drivers, disease markers, and disease-specific survivability-essential genes. Finally, we also included in each network the targets of all
currently available FDA-approved drugs.
To keep our networks and forthcoming analyses focused, we limited the networks to include only nodes at a distance of at most 2 downstream from the targets of FDA-approved drugs, and at a distance of at most 2 upstream from the diseases-specific
nodes of interest. In this way, the networks focus on short signalling paths from the drug targets to disease-specific vulnerabilities. The networks are specific to each patients, being focused on their own genetic abnormalities, and vary
substantially from patient to patient.
The drug synergy data was integrated into the NetControl4BioMed software in the internal search algorithms. The data reflects antagonistic and synergetic relationships between drugs. The search algorithm builds sets of input nodes, preferably
drug targets, that control a given set of disease-specific vulnerabilities. The construction is done step-by-step with one ore more input nodes added in each iteration. For each of the nodes about to be added, the new version of the
algorithm performs an additional check on its synergy between the drugs targeting it and the drugs targeting nodes already selected in the input set.
The simultaneous control of several target sets and networks was analysed as a reduction problem to optimising a single network and a single target set reflecting the original problem. Our solution to the problem is to compose the networks-to-be-controlled
through a merger operation, that essentially is a union between their nodes and edges, with a matching on their common parts.
Stage 3: Validation
01-07/2022
The objective for this stage of the project was to validate the results we obtained. For each of the tumour networks we ran a control analysis to identify control paths originating from nodes targetable with FDA approved drugs and ending in
essential genes specific to the disease. To increase the customisation of the analysis, in some versions of the analyses we reduced the
set of essential genes to those that are also highly expressed in that tumour. To make it possible to validate against multi-drug treatments and standard therapy lines we adapted our control algorithms to identify the 2- and 3-drug combinations
that together control the maximum number of essential genes.
In the case of the multiple myeloma study, 93% of the predictions we made corresponded to a standard therapy lines, while the other 7% corresponded partially (only 2 of the 3 drugs) to one of the therapy lines. Remarkably, in some of the tumours,
the combinations we found corresponded to lines of treatment that are routinely applied later in the disease progression, after the tumour become non-responsive/resistant to other lines of treatment. This suggests possibilities that our
methods may identify ways to combat drug resistance in a personalised way.
In the case of the glioma study, our methods identified a flurry of FDA-approved drugs that may have a significant effect on the network. The much higher number of predictions in this case may reflect the different type of data used in synthesising
the interaction networks: expression data as opposed to mutation data. The concept of simultaneous control, or stratified application of network controllability, had the interesting result of drastically cutting through the likely noise
and resulting in the prediction of only 12 drugs of interest, all of them either standard drugs in glioma, or in clinical study for it. The theoretical efficiency ranking of these drugs varies from tumour to tumour, bringing back the personalised
aspect of our analyses.
SEMINARS
15.7.2022 Project objectives: final evaluation. Final project report. Continuation plans (Ion Petre).
24.6.2022 Drug resistance through interaction networks: final results (Ion Petre). Simultaneous control of several networks: final results (Ion Petre).
3.6.2022 Discussion with Prof. A. Ribeiro on semi-quantitative control of interaction networks. Drug resistance through interaction networks (Ion Petre). PhD thesis disputation (Victor Popescu).
20.5.2022 Simultaneous control of several networks: recent results (Ion Petre). Drug resistance through network modeling (Ion Petre).
29.4.2022 Summary of the discussion with Prof. T. Oprea on networks for personalised oncology (Ion Petre). Validating the glioma predictions (Nicoleta Siminea).
8.4.2022 Software for offline, batch analysis of networks (Victor Popescu). PPI data and its integration (Victor Popescu). Validating the predictions (Nicoleta Siminea).
24.3.2022 The glioma patient networks: current state. (Victor Popescu). The blood-brain barrier and its significance for our predictions (Nicoleta Siminea). The NetControl4BioMed to continue to be hosted on MS Azure (Ion Petre).
10.3.2022 Review of the objectives of the third stage of the project. Plan the next month’s work. (Ion Petre)
17.2.2022 The latest paper published in the project: Victor Popescu, Krishna Kanhaiya, Iulian Năstac, Eugen Czeizler, Ion Petre. Network controllability solutions for computational drug repurposing using genetic algorithms.
https://www.nature.com/articles/s41598-022-05335-3 (Victor Popescu)
9.12.2021 New team member, results on AI modeling (David Pacioianu). Scalable model refinement in biomedicine with Event-B. https://www.nature.com/articles/s41598-022-05308-6 (Ion Petre).
18.11.2021 PhD thesis on network analytics in biomedicine (Victor Popescu). The annual report 2021 (Ion Petre).
14.10.2021 Review of the scientific results of the project so far (Ion Petre). The interaction network for the viral infection model: manuscript accepted for publication (Nicoleta Siminea). The interaction networks for the glioma and
the mieloma case studies: review (Ion Petre, Andrei Paun, Mihaela Paun, Eugen Czeizler, Victor Mitrana, Nicoleta Siminea, Victor Popescu).
16.9.2021 The interaction network for the viral infection model: the referee reports (Ion Petre, Nicoleta Siminea, Victor Popescu, Andrei Paun). NetControl4BioMed: manuscript published (Victor Popescu, Eugen Czeizler).
A new project application (Ion Petre). The personalised interaction networks for the glioma case study (Nicoleta Siminea, Victor Popescu).
24.6.2021 The interaction network for the viral infection model: the manuscript, the poster and the conference presentation (Ion Petre, Nicoleta Siminea, Andrei Paun, Mihaela Paun, Eugen Czeizler). NetControl4BioMed: poster and conference
presentation (Victor Popescu)
10.6.2021 The interaction network for the viral infection model: poster and conference presentation (Nicoleta Siminea). NetControl4BioMed: poster and conference presentation (Victor Popescu).
27.5.2021 The interaction network for the viral infection model: updates (Ion Petre). NetControl4BioMed: updates and conference presentation (Victor Popescu). The personalised interaction networks for the glioma case study (Victor Popescu,
Nicoleta Siminea).
13.5.2021 The interaction network for the viral infection model: updates (Ion Petre). The personalised interaction networks in the glioma case study (Nicoleta Siminea, Victor Popescu). The automatic separation of cancer tissues in glioma
samples (Ion Petre, Andrei Paun).
29.4.2021 Viral infection manuscript and conference abstract. Network topology and Network Controlability for personalised medicine: recent results. (Ion Petre, Eugen Czeizler)
15.4.2021 The glioma interaction networks. Viral infection manuscript and conference abstract. (Ion Petre, Nicoleta Siminea, Andrei
Paun, Victor Popescu)
1.4.2021 The glioma interaction networks. Viral infection manuscript. The next project objectives (Ion Petre, Nicoleta Siminea,
Andrei Paun, Victor Mitrana)
18.3.2021 The multiple myeloma interaction networks. Viral infection network controlability results, drug repurposing predictions
(Ion Petre, Nicoleta Siminea, Andrei Paun)
4.3.2021 The multiple myeloma interaction networks. Viral infection network analysis (Ion Petre, Nicoleta Siminea, Victor Popescu,
Andrei Paun)
11.2.2021 The multiple myeloma interaction networks. Viral infection interaction networks (Ion Petre, Nicoleta Siminea, Victor Popescu, Eugen Czeizler, Andrei Paun, Victor Mitrana, Corina Itcus)
28.1.2021 The objectives for the second part of the project. The SARS-CoV-2 interaction network: from drug targets to essential genes. (Ion Petre, Nicoleta Siminea, Victor Popescu, Corina Itcus)
17.12.2020 Sumarul etapei 1 a proiectului. (Ion Petre)
19.11.2020 The project’s website. Summary of data collection. The annual scientific report. (Ion Petre, Nicoleta Siminea, Eugen Czeizler)
4.11.2020 The project website. The multiple myeloma data. (Ion Petre)
22.10.2020 Drug synergies. The Covid-2019 network: topological analysis, pathway enrichment analysis. (Eugen Czeizler, Nicoleta Siminea, Victor Mitrana)
8.10.2020 The glioblastoma data: patient data, essential genes, treatment lines. The NetControl4BioMed. The Covid-2019 network. (Nicoleta Siminea, Eugen Czeizler, Andrei Paun, Mihaela Paun, Victor Mitrana, Ion Petre, Victor Popescu, Octavian Pacioglu).
24.9.2020 The glioblastoma data: data identification (Nicoleta Siminea)
10.9.2020 Kick-off seminar: project presentation, action plan (Ion Petre)
LATEST PUBLICATIONS
Elio Nushi, Victor Popescu, Jose Angel Sanchez Martin, Sergiu Ivanov, Eugen Czeizler, Ion Petre. Network modeling methods for precision medicine. In: Elisabetta De Maria (Ed.) Systems Biology Modelling and Analysis: Formal Bioinformatics Methods and Tools, Wiley, 2022, to appear.
Nicoleta Siminea, Victor Popescu, Jose Angel Sanchez Martin, Daniela Florea, Georgiana Gavril, Ana-Maria Gheorghe, Corina Itcus, Krishna Kanhaiya, Octavian Pacioglu, Laura Ioana Popa, Romica Trandafir, Maria Iris Tusa, Manuela Sidoroff, Mihaela Paun, Eugen Czeizler, Andrei Paun, Ion Petre. Network analytics for drug repurposing in COVID-19. Briefings in Bioinformatics, 2021.
https://doi.org/10.1093/bib/bbab490.
Victor Popescu, Jose Angel Sanchez-Martin, Daniela Schacherer, Sadra Safadoust, Negin Majidi, Andrei Andronescu, Alexandru Nedea, Diana Ion, Eduard Mititelu, Eugen Czeizler, Ion Petre. NetControl4BioMed: A web-based platform for controllability
analysis of protein-protein interaction networks. Bioinformatics 37 (21), 3976-3978, 2021.
http://10.1093/bioinformatics/btab570.
Adrian Lita, Joel Sjöberg, Stefan Filipescu, Orieta Celiku, Luigia Petre, Mark Gilbert, Houtan Noushmehr, Ion Petre, Mioara Larion, PATH-45. APOLLO: RAMAN-BASED PATHOLOGY OF MALIGNANT GLIOMA,
Neuro-Oncology, Volume 23, Issue Supplement_6, November 2021, Page vi125,
https://doi.org/10.1093/neuonc/noab196.497.
Lukasz Mikulski, Ion Petre. Special issue on Reaction Systems, Theoretical Computer Science, vol. 881, 1-15, 2021.
Usman Sanwal, Thai Son Hoang, Luigia Petre, and Ion Petre. Large Scale Biological Models in Rodin. In: The 9th Rodin User and Developer Workshop, affiliated with the 8th International Conference on Rigurous State Based Methods, 2021.
Nicoleta Siminea, Victor Popescu, Jose Angel Sanchez Martin, Ana-Maria Dobre, Daniela Florea, Georgiana Gavril, Corina Ițcuș, Krishna Kanhaiya, Octavian Pacioglu, Laura Ioana Popa, Romica Trandafir, Maria Iris Tușa, Manuela Sidoroff, Mihaela Păun, Eugen Czeizler, Andrei Păun, Ion Petre. Network controllability analysis for drug repurposing in COVID-19. Accepted for oral presentation at the 29th Conference on Inteligent Systems for Molecular Biology, joint with the 20th European Conference on Computational Biology.
Victor Popescu, Jose Angel Sanchez-Martin, Daniela Schacherer, Sadra Safadoust, Negin Majidi, Andrei Andronescu, Alexandru Nedea, Diana Ion, Eduard Mititelu, Eugen Czeizler, Ion Petre. NetControl4BioMed: A web-based platform for controllability analysis of protein-protein interaction networks. Accepted for oral presentation at the 29th Conference on Inteligent Systems for Molecular Biology, joint with the 20th European Conference on Computational Biology.
Jose Angel Sanchez Martin, Ion Petre. Network controllability analysis of three multiple-myeloma patient genetic mutation datasets. Accepted for poster presentation at the 29th Conference on Inteligent Systems for Molecular Biology, joint with the 20th European Conference on Computational Biology.
Joel Sjöberg, Adrian Lita, Stefan Filipescu, Orieta Celiku, Luigia Petre, Mark R. Gilbert, Houtan Noushmelu, Mioara Larion, Ion Petre. Raman spectra-based agglomerative clustering for tumor detectionin glioma samples. Accepted for poster presentation at the 29th Conference on Inteligent Systems for Molecular Biology, joint with the 20th European Conference on Computational Biology.
José Ramón Sánchez Couso, José Angel Sanchez Martín, Victor Mitrana, Mihaela Păun: Simulations between Three Types of Networks of Splicing Processors. Mathematics
2021, 9(13), 1511;
https://doi.org/10.3390/math9131511
José Angel Sanchez Martín, Victor Mitrana: Simulations between Network Topologies in Networks of Evolutionary Processors Axioms 2021, 10(3), 183;
https://doi.org/10.3390/axioms10030183
Victor Mitrana, José Angel Sanchez Martín: Simulating polarization by random context filters in networks of evolutionary processors. J. Appl. Math. Comput. (2021).
https://doi.org/10.1007/s12190-021-01542-9.
Lukasz Mikulski, Ion Petre. Preface to the Special issue on Reaction Systems, Journal of Membrane Computing, 2: 147-148, 2020.
https://doi.org/10.1007/s41965-020-00047-x
Sergiu Ivanov, Ion Petre. Controllability of reaction systems. Journal of Membrane Computing, 2, 290-302, 2020. https://doi.org/10.1007/s41965-020-00055-x.
Florin Bilbie, Andrei Paun. Small SNQ P Systems with Multiple Types of Spikes. Theoretical Computer Science,862, 14-23, 2020.
https://doi.org/10.1016/j.tcs.2020.10.014
Project results snapshots
Here are some snapshots of the NetControl4BioMed platform that integrates the methods developed in this project. The platform is freely available as web software at https://netcontrol.combio.org/.
The analysis platform
A snapshot of the NetControl4BioMed platform (https://netcontrol.combio.org/).
The data integration flow
A diagram with how the different data sources are integrated in our networks.