‘All models are wrong, some are useful!’ – George Box
The advancement of biological research relies on understanding how the different moving parts interact with each other, forming an integrated system. To better understand a system we often resort to models which help us develop a logical framework of these interactions, offering us predictive insights into how these systems organize and function.
The completion of human genome project brought a new wave of collaborative research in biology. The success of this project was a giant leap towards understanding our genes, the units that code for molecules that conduct the essential functions making human life possible. For the first time the complete sequence of information inside the human genome, the collective set of our genes, was mapped. With the discovery of the sequence of the human genome, efforts were put in to exploit this valuable information for practical insights such as understanding hereditary disorders and genetic basis of deadly conditions like Alzheimer’s disease. Many new exciting disciplines such as bioinformatics and systems biology were developed to tackle and utilize the data generated from the human genome project. These new arenas in science paved a unique way for the intersection of mathematics and biology. For instance, bioinformatics can be used to analyze a large amount of data from biological experiments which will be impractical to be computed manually otherwise. These complex analyses are achieved by developing new mathematical procedures, also called algorithms, and software packages to perform faster calculations. Similarly, systems biology extends a theoretical basis for an observed biological phenomenon. Imagine what Newton did with his theory of gravitation to explain the motion of planets in solar systems but instead of stars, now scientists inspect tiny bacteria under the microscope and try to understand phenomena such as cell division and cell death using mathematics.
One of the powerful ways to model cancer is by using genome-scale metabolic models (GSMM). In principle, genome-scale models are nothing but a mere collection of all the reactions an organism can perform. But, they can be very useful in practice. By using these models, one can compute the rate of all the reactions inside a cell with the knowledge of only a few reaction rates. Genome-scale metabolic models have become powerful tools in recent times to model biology. It has found its use in various areas such as metabolic engineering and synthetic biology. Perhaps, the most successful example of their use is the synthetic production of artemisinin, a powerful anti-malarial drug component. Artemisinin has been traditionally extracted from natural sources and is very expensive but Keasling laboratory genetically engineered yeast cells to produce it cheaply and in abundant quantity. This was made possible by the use of GSMM which helped in designing a precise modification to the yeast which could not have been achieved by a conventional ‘hit and trial’ method which was common among biologists in pre-2000 era. The hit and trial strategy consisted of either guessing potential genes from previous scientific knowledge or conducting large-scale screening of all potential targets until one got lucky. However, this approach is not only time consuming but also economically wasteful. This is where the predictive power of modeling becomes useful. They provide a strategy to make predictions from a limited dataset, thereby saving money as well as improving accuracy.
The applications of GSMM extend to cancer treatment as well. This modeling approach has enabled us to improve our understanding of long observed but poorly understood phenomenon like Warburg effect in cancer. GSMMs have also made it simpler to identify essential genes, whose inactivation kills the cells. Predicting these genetic targets from the model drastically reduces the number of options which need to be screened to kill cancer cells. This is of much therapeutic interest because of the expenses and labor associated with large-scale screenings. However, GSMM can make the job very simple and can provide predictions in a few hours using fast algorithms. In fact, GSMM for cancer has proved useful in identifying essential genes in cancer which were previously unknown and since then have been experimentally verified. GSMM can also provide functional insights into the genetic function at the same time.
However, one needs to be careful while modeling biology. Biology is complex with multiple processes happening parallelly that are difficult to monitor. The assumptions in the model are not always realistic and thus modeling results may offer many positive predictions which turn out to be false and vice versa. Indeed, all results are hypothetical unless they have been experimentally proved inside labs. However, genome-scale models are still useful because they offer predictions which can easily be tested and compared to lab-based results. To top it off, GSMM is very easy to use, and any biologist with minimal understanding of mathematics can use them as a stepping stone to guide their experimental study. By exploiting the predictive capabilities of GSMM, it is possible to streamline experimental endeavors in cancer biology, and we hope the insights we gain from modeling can be translated into therapeutically relevant findings.