After getting inspired by looking at Nathan Yau’s website, I spent some time today playing with data visualization. I pulled some data from the awesome NYC OpenData website, which I plan to grab a ton more from, and stuck it into MATLAB to do some analysis / plotting, and finally threw an exported PDF into illustrator for some very light editing and text addition. Click the image above for a PDF if you’d like to mess with it or use it in any way.
Imagine a machine designed to slice bread which, through some pathological design concept, posessed the trait that its blade was also somehow its power source. Removing the blade/power supply would clearly render the device inoperable, but understanding how this action had achieved its effect would be quite difficult. This is the essence one of the main problems which confronts anyone interested in teasing apart the complex web of interactions that is molecular biology.
For whatever reason, whether it be a basic feature of human intelligence or simply a sort of paradigmatic immaturity as a species, we tend not to design things in the same way the evolution does. By that I mean employing multi-purpose parts in the way the fictional device mentioned above does. One human-designed object posessing that property is the bicycle. There, it so happens, that the rotation of the wheels actually tends to keep the bike up-right. The wheels are like gyroscopes: their rotational intertia tends to keep them in their plane of rotation in the same way that the linear inertia of an object moving in a straight line tends to keep it on that trajectory. In the sense that the idea of the bicycle has retained this accidental advantage, it resembles an evolution-designed object. However, examples of human engineering that fit this category are few.
In biology, it appears, this type of overlapping, redundant functionality is the norm. For example, insulin is a molecule which is well known to many layman as being involved in the metabolism of glucose: the regulation of blood sugar. However, if one simply consults the wikipedia article on insulin, it immediately becomes clear that this is far to simple a tale. The functions of insulin listed there are:
1. Increased glycogen synthesis – insulin forces storage of glucose in liver (and muscle) cells in the form of glycogen; lowered levels of insulin cause liver cells to convert glycogen to glucose and excrete it into the blood. This is the clinical action of insulin which is directly useful in reducing high blood glucose levels as in diabetes.
2. Increased fatty acid synthesis – insulin forces fat cells to take in blood lipids which are converted to triglycerides; lack of insulin causes the reverse.
3. Increased esterification of fatty acids – forces adipose tissue to make fats (i.e., triglycerides) from fatty acid esters; lack of insulin causes the reverse.
4. Decreased proteinolysis – forces reduction of protein degradation; lack of insulin increases protein degradation.
5. Decreased lipolysis – forces reduction in conversion of fat cell lipid stores into blood fatty acids; lack of insulin causes the reverse.
6. Decreased gluconeogenesis – decreases production of glucose from non-sugar substrates, primarily in the liver (remember, the vast majority of endogenous insulin arriving at the liver never leaves the liver) ; lack of insulin causes glucose production from assorted substrates in the liver and elsewhere.
7. Increased amino acid uptake – forces cells to absorb circulating amino acids; lack of insulin inhibits absorption.
8. Increased potassium uptake – forces cells to absorb serum potassium; lack of insulin inhibits absorption.
9. Arterial muscle tone – forces arterial wall muscle to relax, increasing blood flow, especially in micro arteries; lack of insulin reduces flow by allowing these muscles to contract.
Even as I’m writing this, I have come across an article in Nature about a previously unknown action of insulin in a biochemical pathway involving a protein called TORC22.
Some would say that I am pointing to an inherent flaw in reductionist thinking. That our tendency to search for the smallest parts in order to build up a description of everything from the universe itself to the many varied forms of matter we find within it, cannot hope to penetrate these massively interconnected systems. It seems true that our current notions of what the smallest parts are will lead us to descriptions which are simply too large scale to be intuitively understood. However, this doesn’t necessarily point to a flaw in reductionism, especially since the alternative approach of holism doesn’t seem to offer any ways forward which circumnavigate such a problem. Rather, I would suggest that we need to fundamentally shift the way we think about evolved things in order to make significant progress towards understanding that which falls under the blanket term of “complex systems.”
Early in the last century, physics was spurred on by a shift in thinking: quantum theory, often thought of as one of the canonical scientific revolutions. I am hopeful that this century, or some time in mankind’s future, will do the same for biology, and complexity in general.
1. Durrett, Rick. Random Graph Dynamics New York: Cambridge University Press, 2006.
2. Dentin, R. Liu, Y., Koo, S.H., Hendrick, S., Vargas, T., Heredia, J., Yates, J. III, Montiminy, M. (2007) Insulin Modulated gluconeogenesis by inhibition of the coactivator TORC2 Nature 449: 366-369