Date

2017-05-10

Author

Karakelle, Barış Su
Purutçuoğlu Gazi, Vilda
Ürün, Yüksel

Metadata

Show full item record

Item Usage Stats

231
views

0
downloads

Cite This

The cancer disease is the second most common disease type seen after the frequency of the cardiovascular diseases. The frequency of this genetic disease changes with respect to the gender. Accordingly, the gynecological cancer, which covers ovarian, endometrial or cervical cancer, is the second most common cancer type in women after the breast cancer. Similar to other cancer types, the gynecological cancer is the system disease, meaning that the malfunctions and mutations in the gene regulatory pathways cause this illness. Hereby, in order to diagnoseand treat it, it is crucial to understand the activations and the crosstalk of the biological systems that are affected by this cancer. In this study, to detect the changes in interactions between the genes of the same and the distinct pathways under the gynecological cancer, we comprehensively check the associated literature and make a list of the most affected genes. Then, we consider that these core genes can be represented artificially as an oncogenic network and estimate the strength of their interactions by a mathematical model. Later, we extend our system by adding other proteins within the most affected two regulatory networks, so-called MAPK/ERK and PI3K/AKT pathways, under this illness. In this case, the total number of genes used in the analysis becomes 41 proteins. After we infer the interactions of this complex structure by a mathematical model, we further extend our system by including the proteins affected by the crosstalk between systems resulting in 120 proteins roughly. Finally, we estimatethis largest network. In all these analyses, we use a microarray dataset which is composed of 3171 genes that are collected under various cancer types. On the other hand, in the underlying three-level inference, we apply the Gaussian graphical model in the mathematical description of the systems and estimate the model parameters via the graphical lasso approach. From the first stage of modeling, we obtain promising results which can be fully validated by the knowledge in the databases. For the following two models, part of our findings are validated and the remainings are reported as an interesting list of interactions that can be further studies. As the future work, we consider to combine different datasets to describe a more general view of this oncogenicnetwork and investigate other mathematical models.

Subject Keywords

Cancer modeling, Gaussian graphical model, Biological networks, Systems biology

URI

https://hdl.handle.net/11511/75829

Collections

Department of Statistics, Conference / Seminar