by
lmh » Fri May 06, 2011 1:22 pm
Peter, can I just chip in?
(1) When I first read your question, my immediate thought was "great, that's the sort of question Peter Apps will be able to answer for this chap"...
(2) I'd be a bit wary, from a psychological point of view, of what you're trying to do. If I were the peer reviewer, my thoughts would be along the following lines: "Fine, it's theoretically possible to have a situation where every individual is identified by a single chemical totally unique and found in no other, but how is this achieved biochemically? Are you expecting me to believe that every individual has made up some chemical pathway absent from every other? That every individual has a unique synthetic pathway. This sounds like a very un-Occam's-razor approach, an unnecessarily complicated hypothesis. Nevertheless, you're right to question whether we should set ourselves up in a way that prevents us from noticing if it's true. But is this a case of an analytical expert arguing for bigger and better equipment and more money for developing methods that are more sophisticated than we really need (arguments like "Biology is complex, so give me more money please" ring alarm bells for me)? Surely if every individual has their own chemical, we'll notice pretty soon that we're seeing novel peaks, even if we can't actually separate every novel peak. We don't need a method that will separate 4 billion chemicals, we just need a method that has enough blank space between its peaks to provide a 'trap' in which to catch a new chemical, if one exists."
In effect, I'd regard this very large number as sort-of-interesting but not really changing how I felt about a paper. I'd be more alarmed by the dog's-nose-LOD issue.
(3) Practically I don't think you can answer your question. You can estimate how many chemicals are theoretically possible, as you have tried, but you can't estimate how good we need to make our method if we wish to see all chemicals that really exist, because we only know they really exist if we have a method that can see them. It's currently a circular argument.