Improving Drug Delivery using Network Analysis
An interesting application of networks and graph theory is the improvement of drug delivery systems and locations. A short lecture given by Ernest Fraenkel (sourced below) touches on the intersection between system biology and network systems. Cells, proteins, genes, chemicals, and other biological components can potentially be as viewed as nodes, while their interactions–for example, expression of a gene due to the presence of a hormone, or communication traveling from a cell receptor to protein to nucleus, which ultimately leads to some physiological response–is considered an edge. This model of proteins/genes as nodes and edges as potential physical interactions between nodes as the “Prize-collecting Steiner tree.” The problem with this approach is that finding significant interactions amid the countless reactions occurring in a biological system becomes extremely difficult.
The advantage of viewing biological interactions as a network is being able to see significant nodes that were not recognized as important through experimental data alone. On the other hand, irrelevant nodes can be eliminated, which creates a sub-network within the larger biological network. This sub-network’s context can be used to guess the function of the nodes in that sub-network. For example, this process could potentially be used to find important genes involved in a disease. Finding a node that is central in its network could be used to improve drug delivery and alter processes more effectively.
A more concrete example: Glioblastoma is a brain tumor usually diagnosed at a late stage, and combated with chemotherapy and surgery. This disease is modeled by stating that a gene that, when mutated, plays a prominent role in the growth of the tumor, has mutated; the next step is to then look at the effect of this gene on the rest of the cell. The network system developed from this modelling shows genes that play a very important role in the behavior of the cell post-mutation that were not found experimentally. The SRC gene especially was found to be very central in the network of genes, and it was found experimentally that targeting SRC and the mutation simultaneously is effective at slowing the growth of the tumor. A node (a gene called ESR1) connected to SRC but not the mutated gene was found experimentally to be irrelevant. However, using the network model, it was found to have an even stronger effect on the growth of the cell when drug delivery targeted the mutated gene, SRC, and also ESR1. This relationship makes sense when ideas learned in lecture are applied. If SRC and the mutated gene are found to have a strong connection, and SRC and ESR1 have a strong connection, through the strong triadic closure property, it makes sense that the mutated gene and ESR1 also have a relationship.
Source (video): Systems Biology: Where Computer Science, Engineering and Biology Meet