Part of today's complex festivities included a presentation about networked reactions to drugs.

It was chilling.  (Speaking of chilling, don't get me started on medical errors.  Did you know that there are between 50,000 and 200,000 deaths per year in the US that are caused by straight-out avoidable errors?  Wrong drug, wrong leg cut off, wrong treatment.  If there are that many deaths, just imagine how many mistakes are made that don't lead to death.)

The problem with administering medications is that 70%-90% of proteins communicate with one another.  So if you do something that you think is inhibiting X protein, you may also (in some downstream reaction) be cheering up Y protein, and thus causing the exact opposite effect of what you wanted in the first place.  This is the Vioxx story.

There are lots of levels of this, from simple double-edged sword effects -- like the conflicting downstream effects -- to feedback loops/synergistic effects in which there are two or more stable states in which two drugs can coexist.  If you want to drive the system to one of those states (to make the patient well), you may have to find a third drug that will get you there. 

Or you could try to map incredibly complicated genomic interactions with drugs to figure out how to control the fate of various cells.  Sure -- 13,000 genes, 32 drugs -- just draw the picture.  With some funding assistance.

These are all very brave and optimistic efforts to figure out what a highly-connected network does when medications are added in.  The lesson I took away was this:

We really don't know whether any particular drug will fix any particular problem, or whether it will instead/also cause a different problem down the road. 

Whenever we try to manipulate a complex system, we're shooting bullets in the dark.