BBSRC Systems Biology Grantholder Workshop, University of Nottingham, 15 December 2008.
He obviously agreed with my thoughts on outreach: he started his talk describing his recent outreach activities. He spoke recently at EMBL Heidelburg, Quebec, Paris, Bejing, Bangkok (buddhist monks!).
FIRST PRINCIPLE: biological functionality is multi-level. Genes do nothing on their own. they are simply databases (no "genetic program"). Physiological functions use many genes in collaboration. Determining the level at which a function is integrated is one of the aims of Systems and Computational Biology. One experiment (in his book) concerns competing metaphors: given the statement "genes as prisoners verus selfish genes", what experiment could possibly distinguish between the two diametrically opposite metaphors? He then said that Dawkins is a "beautiful writer", but simultaneously "catastrophically wrong". It was a line that got a laugh from the general populace. Dawkins wrote that "an organism is a tool of DNA rather than the other way round". How can you distinguish between this selfish gene statement, and the prisoner concept, where genes are "trapped in huge colonies, locked inside highly intelligent beings, moulded by the outside world, communicating with it by complex processes, through which, blindly, as if by magic, function emerges." (The Music of Life, chapter 1). Both ideas are metaphorical, and there are ABSOLUTELY NO EMPIRICAL TESTS THAT YOU CAN DO TO DISTINGUISH BETWEEN THEM. He feels that it's politics, not biology.
He then provides an illustrative calculation of gene knockouts: assume (absurd, but instructive) that each function depends on 2 genes. Then, the total number of posible functions would be 0.5 x 25,000 x 24,999 = 312,487,500. This is even larger if you assume more realistic numbers of genes involved in each "function". This leads him to the second principle.
SECOND PRINCIPLE: Transmission of information is NOT one-way. So, the "central dogma" of biology is incorrect! There is downard causation from all levels. This influences gene expression and gene marking (epigenetic inheritance – Qiu 2006, Nature 441). "Lamarkism is not so obviously false as is sometimes made out" (John Maynard Smith, Evolutionary Genetics, OUP, 1998). His suggestion to unravel complexity is that you need to work in an integrative way at all levels, and in both an upward (gene-> protein->pathways->etc) and downward (systems-level controls of gene expression, protein machinery reads genes, systems level triggers of cell signalling, epigenetic marking by all levels) direction. A lot of the feedback is constraint, rather than actual feedback. This leads to the third principle.
THIRD PRINCIPLE: DNA is NOT the sole transmitter of inheritance. E.g. DNA methylation, histone marking/modification, and other processes. Example in the press: The Guardian, 14 February 2007, "Motherly love may alter genes for the better." Paper is in the Journal of Neuroscience, 27(7):1756-1768. In colonies without predators they do more grooming. Without predators they do less. Both of these behaviours are continued in the next generation: could think it's transmitted culturally. However, it turns out is transmitted epigenetically. It is a form of aquired, imposed behaviour that jumps the germ line.
FOURTH PRINCIPLE: The theory of (biological) relativity. There is no privileged level of causality in biological systems (and a multi-level analysis is therefore necessary).
FIFTH PRINCIPLE: The Gene Ontology will fail without higher-level insight into what a gene actually is. Most genes are ancient. You can use a linguistic metaphor fairly successfully (Genes and Causations – his article where he makes his own definition of a gene.) (Personal opinion insert: I think that another way of stating this might be that ontologies in general will fail unless they are well thought-out and well constructed. I'm not sure, but he may have been conflating gene *names* with gene *terms* in an ontology during the talk, but that might have been me mishearing.)
SIXTH PRINCIPLE: There is no "genetic program" (term invented by Monod and Jacob). DNA as a tape that can be fed into a computer. Noble doesn't see a program there – it's not a separate set of code – everything (functionality, etc) is present in the sequence.
SEVENTH PRINCIPLE: There are no programs at any other level (not at the genetic level, nor any other level). References Sydney Brenner's "middle-out" method of research (as opposed to top-down or bottom-up).
Noble then told us about he started modelling years ago, and what types of computers were in use originally. (Huxley took 8 hours of grinding on a Mrunsviga Model 20 to get 5 milliseconds of data!) When Noble played table tennis with Huxley, Huxley won easily and Noble believes it was due to all that exercise on the computer! Another machine played a song, "Oh dear, what can the matter be", when something went wrong (didn't catch which one, sorry!). He then described an example of protein interaction in a cell model: how to reconstruct the heart's pacemaker.
Reverse engineering to obtain absolute values for gene contributions to function (Noble 2008, Genes and Causation, Phil Trans Roy Soc A, 366, 3001-3015). Functions are generally robust: one gene being knocked out might just cause another's contribution on the functionality to increase. Therefore measuring contributions of individual genes is very difficult. Noble believes that reverse engineering could solve this problem. Many systems beautifully buffer knockouts (the example he gave is the model of the sinus node). What he does is measure at a higher level the contribution of each component to the electric current. Therefore you can reverse engineer from the model, e.g. this component provides 80% of the depolarizing current. Further, by discovering such a buffering mechanism, you can create a safe treatment for angina, because you can use those components which are buffered as drug targets! (From this, got ivabradine for the treatment of angina. This works by restraining the frequency and particularly limit the extent to which the frequency increases after adrenalin.)
From these and similar examples, he is showing that SB is most definitely delivering the goods. What we need to do next is to connect levels, or the incorporation of cellular models into organ or organism level models. With the Auckland model ventricle, you combine the engineering + electrical + biochemical models! Showed a fantastic video of heart fibrillation by Panfilov and Witkowsky.
There are some other people working on a higher-resolution model in Oxford using images from a heart, which shows in very fine detail the collision of the waves of starting from the apex of a heart. Takes a while to run, so they've now got a contract with Fujitsu to run whole-organ models, which should allow them to run faster than real-time.
EIGHTH PRINCIPLE: No programs at any level – including the brain!
NINTH PRINCIPLE: missed it – urgh! (Update: Simon's provided it (see comments): "the self is not an object. The mind is not a separate object competing
for activity and influence with the molecules of the body." And, now that he's written that, one of Denis' comments makes sense – while describing the 9th principle, he mentioned that it was this one that got the buddhist monks interested in his talks).
TENTH PRINCIPLE: There are many more to be discovered. The theory(ies) of biology does(do) not yet exist. (He doesn't regard evolution as a theory, but a fact which stands in need of theories.) Further, biology here, even if it were all explained, wouldn't necessarily hold on other planets 🙂 That is the challenge for systems biology: trying to figure out how to understand biology.
The 10 principles are outlined in the Paton Lecture 2008, in Experimental Physiology 93 16-26.
Noble doesn't use the word "emergence" for much the same reason he wonders whether he should use the word "feedback". He recommends Kupiec's new book (Origin of Individuals, just coming out, already out in France), which deals with the interaction between genes and environment in which various expression levels develop. DNA isn't a blueprint for the organism (when you might be able to talk about emergence): instead, you should describe an organism as an interaction machine, and not an emergence machine (where "emergence" refers to a map that's being unfolded or unravelled).
The challenge to the mathematicians is to determine what kind of maths is required for connecting the different levels of granularity of models
Thanks very much to Denis Noble!
These are just my notes and are not guaranteed to be correct. Please feel free to let me know about any errors, which are all my fault and not the fault of the speaker. 🙂
Update: Just noticed Phil's post on the same talk.