Maybe I am a bit biased, being a mathematician and all, but math tends to be absolutely everywhere in science. Be it economics, physics, chemistry, computer science, earth science or as you may have guess by the name of the article, biochemistry. One of many strengths of mathematics lies in its ability to model complex behaviours and predict outcomes without performing thousands upon thousands of physical experiments. Using math one can “simply” model the system, throw it in a computer program and let it do the heavy work by pressing a button. Out comes a predicament which you then can use to adjust your real system or experiment. This process can take less than a few hours and give instant results, making it very efficient compared to old school laboratory work.
The “simple” math and flux balance analysis in a nutshell
But okay, so if you’re reading this and have no idea of what I’m talking about, I might oversimplify the process just a tiny, tiny bit. Especially regarding the efforts needed to build the programs and create the models. This often proofs to be a very difficult and time-consuming process and the models created are usually very complex and hard to solve or optimise, since reality often is very complex. This of course leads to having to do necessary approximations and be satisfied with sub-optimal solutions. Scientists often aim to simplify the model with assumptions and approximations, though as few as possible, making easy-to-solve equation systems. This, while still aiming to retain the essential information of the system and predictability of the model.
For metabolic networks this is exactly what flux balance analysis (FBA) aims to do; assuming the studied biological system is in steady state and that the biological system has reached some kind of evolutionary goal (don’t dwell on this, when you use FBA yourself you know what this means), a metabolic network can be modelled and solved using simple linear systems and linear algebra.
You don’t know linear algebra? Don’t worry, all you need to know is that as a mathematician this is where I show up and use my skills to shine; linear equations or linear optimisation problems are well studied, and efficient solution strategies have been developed and implemented for a long time. This is where we mathematicians can push the biologists to the side and proudly take over effortlessly trying to look cool. Or at least we could until some bright person makes a simple program for biologists to use, yet again rendering us unnecessary. I’m not bitter, no not at all …
You don’t need to know programming
I have several biologist friends who always complains about how they hate programming and can’t get rid of all errors even if their life depended on it. Good news my friends, there are countless of pre-developed programs solving these types of models aimed for scientist with minimum programming experience, making it easier to progress in studies and progressing research further. For the interested, Karthik R. and Nagasuma C. [2] outlines several programmes that can be used for solving FBAs. But I’m not going to bore you with such stuff, neither am I going to explain how to build an FBA model. If you want to learn that you will have to do like the rest of us, use Wikipedia and some random video on YouTube. Instead, I want the rest of the text to focus more on some areas of application where FBA CAN and HAVE been used.
Before I forget explaining FBA again …
Before that, let’s first un-mystify the words making up flux balance analysis. Flux here refers to a metabolic flux, e.g., turnover of molecules through a pathway. Balance refers to mass balance constraints on the system and finally, we are analysing it, so analysis fits as the last word. Basically, putting it together, we are calculating the flow of metabolites in a biological system. We can use this to predict for example the growth rate of organisms or production of metabolites which directly throws us into the application discussion.
FBA in practice
So how can we use this in real life? Two sort of obvious applications would be analysis of genome-scale metabolic models, and analysis of metabolic capabilities since they are directly related to the equations at hand. But, as always, there are several more. Here’s two of them:
Drug target identification
A more, according to me, interesting application would be for instance drug target identification. In some cells, certain malfunctioning enzymes produce excessive amounts of certain compounds leading to excessive concentration or mass flow. In drug target identification, the idea is to identify enzymes which can be manipulated using drugs to adjust the production of such compounds. So how can we use FBA to find these enzymes? One way is to use so called In Silico (fancy word for computer simulated) gene deletion, where you delete a specific enzyme and study the flux changes (FBA!!) in the system due to the missing enzyme. If the flux changes are large, then the enzyme has great effect on the system, and we can start experimenting with this enzyme. This might be a rather simplistic view of the process, but yeah, something like this. FBA can of course be made more complex and be used to do more advanced things. For instance, Zhenping et al. in addition calculated the drug dose needed for curing certain disease, without ever testing it on any animals or humans. So, in silico models and FBA can help us move away from animal testing. Cool right!!??
Bioremediation in marine food
Another cool application area which Marianna et al. studied is bioremediation. To be more specific; bioremediation in marine food webs. For those who don’t know, bioremediation is basically when you have contaminated, toxic or polluted environment and use organisms like microbes to remove it. So, what Marianna et al. basically did was that they created an FBA model (also ecological modelling but let’s not dive into that) and then they studied In Silico the effect of different methods to remove contamination and what effect it had on the marine life. They did not need any contaminated test water and they did not need to have marine life to test the methods. All they needed were to create a mathematical model, get a computer, use a program to run the model and a button to run the simulation. Poof, bang and done!!! The results which they reported, for those who are interested, were that they deemed their framework to be a good practical tool for simulating sea scenarios and they even compared their results to a real test from the Adriac sea, displaying many similarities.
Good to mention is that this is not always the case. As previously mentioned, creating models can be hard and it can be difficult to determine if it is wrong. Thus, even if it seems like In Silico methods like FBA would bring nothing but benefits, there is still a human making the model and if that human take a wrong step, then the model is useless and maybe even misleading. Oh well.
Last note
What I want you to take away from this post is that math is cool and even “simple” math can be really powerful. Biochemistry is cool, but not as cool as math. Also, you don’t need to know programming, let us programmers deal with that and do what you like in life. In Silico is a cool buzzword, use it to look smart.
Cheers,
Robin
[1] Andre Wegner, Johannes Meiser, Daniel Weindl and Karsten Hiller, How metabolites modulate metabolic flux. Current Opinion in Biotechnology, volume 34. 2015,
[2] Karthik Raman and Nagasuma Chandra, flux balance analysis for biological systems: application and challenges., Briefings in Bioinformatics, 15 mars 2009
[3] Zhenping Li et al., Two-stage flux balance analysis of metabolic networks for drug target identification, BMC Syst Biol. June 20 2011
[4] Marianna Taffi et al., Bioremediation in marine ecosystems: a computational study combining ecological modeling and flux balance analysis, frontiers in Genetics, 12 Sept 2014
