Network Analysis of Biological Systems
Limitations in quantitative analysis of complex biomolecular network outcomes is a major barrier to the treatment of many diseases. In fact, most biological hypothesise can only be tested through cumbersome, expensive and slow in vitro techniques such as the Kirby-Bauer test. This project
2025-06-28 16:34:15 - Adil Khan
Network Analysis of Biological Systems
Project Area of Specialization Artificial IntelligenceProject SummaryLimitations in quantitative analysis of complex biomolecular network outcomes is a major barrier to the treatment of many diseases. In fact, most biological hypothesise can only be tested through cumbersome, expensive and slow in vitro techniques such as the Kirby-Bauer test. This project aims to develop, verify and compare network analysis techniques for biological systems, mainly gene regulatory networks. Gene regulatory networks represent how various genes respond to environment and affect each other in cells. Such networks are abstracted as directed graphs that transition to minima (attractors) of an energetic landscape as postulated by Waddington in his epigenetic landscape. Network analysis techniques may, hypothetically, enable us to predict final attractor(s) (in terms of gene expression profiles) the cell takes and the trajectory to it given a set of initial conditions. However, incomplete information regarding interaction parameters and high complexity of chemical rate equations prevent such calculations. For that, we intend to adapt various network analysis techniques and test their applicability on biological systems followed by estimation of their accuracy in comparison with the empirical data. Experimental data will be obtained from public databases and curated for testing requirements. Analysis techniques will be adapted from a variety of related fields such as Graph theory, Chemical Reaction Network Theory and Stochastic Differential Equations. These will be followed by comparison through intermediate functions such as clustering. Summarily, we will provide a framework for verifying and benchmarking the applicability of various analysis techniques to biological systems towards a deeper understanding and control of underlying cellular dynamics.
Project Objectives- To obtain empirical data of gene expression profiles from public databases and process it to suit testing requirements
- To test existing methods of analysis such as Deterministic Analysis (DA), Probabilistic Analysis (PA) and Ordinary Differential Equation (ODE) modelling
- Adapting and developing set of techniques from related or analogous fields to test applicability
- Developing tools and standards for benchmarking of various methods
- Once accurate and reliable techniques have been identified, to integrate approved drug databases to predict sets of drugs to induce reversion (to normal growth) or apoptosis (cell death) in cancerous cells
A database containing gene expression data across a number of cell lines under varying conditions will be appraised and selected. This will be done based on experimental technique, range of data, reliability and ability to cater to information gaps (such as genome sequencing data for mutations).
Methods of Analysis will be implemented into MATLAB. Lengthy literature reviews will be undertaken to search for mathematical frameworks that may cater to our requirements. These techniques will then be adjusted, as appropriate, before implementation in matlab.
Drug information will be taken from any public database of drugs and their target genes/ nodes, such as lists of FDA approved drugs, and incorporated into our formalism to predict specific drugs for various cell lines.
Benefits of the ProjectThe main benefit will be identification of targets for drug therapy in various diseases, particularly cancer. Once a network analysis technique is proven to be more accurate, it is possible to identify genes that determine cell fate exit from one attractor state to another, such as from uncontrolled division (as in cancer) to apoptosis (cell suicide), effectively allowing targeted therapy based on patient gene mutation panels. The application need not be restricted to cancer as various diseases may be analyzed by this technique.
Moreover, with the ability to simulate biological systems in silico, a variety of experimental hypothesis regarding cell mechanics may be developed for verification. Hypothesis regarding the structure of GRN’s, development trajectories and other phenomenon may be tested physically; in short permitting a targeted, quantitative study of biological systems for research purposes in related fields such as cellular biology, neuroscience, and synthetic biology etc.
Technical Details of Final DeliverableA software prototype, that takes gene expression data as input (along with supporting information such as mutation/ sequencing data, if available) and produces lists of attractors or final states that the cell may take. After confirmation, it predicts the least invasive drug combinations as a form of personalised cancer therapy.
Final Deliverable of the Project Software SystemType of Industry Health Technologies Big DataSustainable Development Goals Good Health and Well-Being for PeopleRequired Resources| Item Name | Type | No. of Units | Per Unit Cost (in Rs) | Total (in Rs) |
|---|---|---|---|---|
| Total in (Rs) | 80000 | |||
| Hard disk & GPUs | Equipment | 2 | 35000 | 70000 |
| Stationary & Printer Accessories | Miscellaneous | 1 | 10000 | 10000 |