Development and application of concepts and methods of complex network theory for characterization, modeling and simulation of biological processes in medicine.

Complex network theory is a highly interdisciplinary field with applications in sociology, physics, engineering, systems biology and medicine. Particularly, in the case of medicine, the network modeling enables the analysis of diseases from genetic to social level. Several works have showed that some diseases are connected at the genetic level. For instance, subjects having one copy of the allele of the gene FTO have an increase of 30% in the likelihood of obesity, increasing to 67% if two alleles are present. Genes that cause obesity are also associated with other diseases, such as diabetes and lipodystrophy. In this way, such diseases can have a common genetic origin. Thus, human diseases are not independent, but form a complex networks in which two diseases are connected if they share at least one gene.

In the social level, social relationships, such as familiar ties and friendship, can influence the spreading of obesity and infectious diseases. When two subjects are friends, if one of them becomes obese, the other had a 171% increased chance of becoming obese too. Furthermore, infectious diseases, such as sexual or respiratory diseases, spread from person to person via social contacts. In this case, it is important to quantify how the structure of social networks influences the propagation in order to develop procedures to control the disease spreading and implement new vaccination strategies.

Some diseases are related to the phenotype. For example, the organization of our brain is associated with neuropsychiatric disorders, such as schizophrenia and autism. Particularly, in a recent paper, we analyzed cortical networks obtained from subjects presenting child-onset schizophrenia, which is a rare form of schizophrenia having its onset before age of 13. From first medical consultation to correct diagnosis of child-onset schizophrenia, it can take years partially because the diagnostic challenges are with differentiating childhood-onset schizophrenia from other neurological disorders or acute, transient syndromes. The data considered were extracted by functional magnetic resonance imaging (fMRI) applied in the brain of patients with child-onset schizophrenia and healthy subjects. Among 54 network measures tested, we verified that only four contributed substantially to a discrimination between the classes of healthy and schizophrenic subjects, with a sensitivity of 90% and specificity of 74%. However, such classes of networks did not differ significantly with respect to the level of network resilience and synchronization. Therefore, the structural differences in the brain of healthy and schizophrenic patients are very subtle. Our findings suggest that it is possible to establish data analysis routines that allow automatic diagnosis of a multifaceted disease like child-onset schizophrenia based on fMRI data of individual subjects and extracted network properties.

In this part of the project, we aim at the characterization, modeling and simulation of biological processes in medicine. Methods for automatic diagnosis of neuropsychiatric disorders, such as schizophrenia and autism, based on the characterization of cortical networks and data mining concepts, will be implemented. In addition, new models of epidemic and obesity spreading will be developed. Finally, we will characterize and analyze protein-protein interaction and transcriptional regulatory networks to determine the genes associated with different diseases and construct the disease network. The collaboration with the research groups at INCOR will be important for the development of the project, since they have experience in analysis of medical data.

Activities

1. Automatic diagnosis of neuropsychiatric disorders. We aim at developing methods for automatic diagnosis of neuropsychiatric disorders based on concepts of complex network theory and data mining. Data of subjects presenting schizophrenia, autism and epilepsy will be considered. These data will be collected from different sources and the analysis will complement our previous works.
2. Modeling of epidemic spreading. Infectious diseases propagate through the contact between healthy and infected individuals, defining a complex network of social ties. Understand how the network organization influences the propagation of pathogens is a fundamental issue in computational epidemiology. Another challenge is the development of more realistic models, which can take into account dynamical interactions or connections in different layers, in order to represent familiar connections or friendship. In this project, we aim at the quantification of the influence of network structure on the spreading of diseases and the developing of more realist models, which can represent the dynamics of pathogen propagation with more accuracy than the current models. These models should consider stochastic fluctuations, time delay, seasonal components and heterogeneous connections. The model validation will be performed by the comparison with real data, such as in the case of the H1N1 propagation. Methods for efficient vaccination of subjects will also be implemented and compared with traditional approaches. These methods should consider only local information, compartmental organization and multi-layer networks. Different types of diseases will be analyzed and the modeling will be performed according to the pathogen properties.
3. Systems biology: The reductionism has been considered during several centuries to understand the cellular components by relating elements, such as genes and proteins, to specific biological functions. However, despite the success obtained, great part of cellular functions cannot be described by the simple analysis of individual components. In this way, it is fundamental to study the structure and dynamics of cell constituents to understand the complex mechanisms involved in cell behavior. The cell activity can be divided in three basic levels: (i) metabolic, which is determined by (ii) the protein interaction, whose production is controlled by (iii) genetic interactions. Thus, to understand the biological processes that control the cell functioning, it is necessary to analyze how the energy is obtained by the cell through biochemical reactions; how proteins participate is several biological processes, such as protein complexes; and how the information are transferred from DNA to mRNA. These systems can be modeled as networks, since they are determined by discrete components (molecules, genes or proteins) that interact. In this part of the project, we aim at the analysis of transcriptional regulatory and protein-protein interaction networks in order to identify how protein and genes associated with different diseases are related. Methods of statistical inference will be considered to determine how accurate the current databases are. Furthermore, data mining techniques will be used to predict the function of proteins and genes from the network structure. Transcriptional regulatory networks will be considered to construct the network of diseases. In this case, two diseases that are related to at least one gene are connected. These studies will enable the development of probabilistic models to predict the occurrence of diseases from the network topology. In addition, an analysis of modular organization of networks will be performed to determine whether diseases with similar properties are related to genes placed at the same community.

Goals

The research proposed in this part of the project will enable us to reach the following goals:

1. Up to 2021: Understand how diseases spread in different complex networks and determine how different properties of epidemic models, such as transmission rate or time delay, influence the pathogen propagation.
2. Up to 2021: Develop more comprehensive models of epidemic spreading.
3. Up to 2021: Develop methods for automatic diagnosis of neuropsychiatric disorders based on concepts of complex networks and data mining.
4. Up to 2021: Analyze protein interaction networks and transcriptional regulatory networks in order to determine how diseases are related at the genetic level and determine new methods for disease prediction from the structure of the disease networks.

Impact

Our research will help to understand how the network organization influences the propagation of diseases. This study will enable the development of methods for control of epidemic outbreaks. New models of epidemic spreading will also be implemented. In addition, several recent works have demonstrated that neuropsychiatric disorders are related to the brain organization. In this way, our research aims at the development of new methods for automatic diagnosis of neuropsychiatric diseases, such as schizophrenia and autism. In the genetic level, the analysis of transcriptional regulatory networks and protein interaction networks can enable to understand how diseases are connected and develop new methods for disease prediction. Since all investigation proposed in this part of the project are new, it is expected that these studies will yield several papers that will be submitted for publication in international journals of computer science, physics and medicine. Furthermore, some PhD students will be involved in the works proposed, which will help in their graduating projects. These works should be developed in collaboration with researchers from INCOR.