How good are we at predictions in science?

During Jon Butterworth's presentation at the UTC on Friday, I kept on thinking back to a comment from the late Richard Feynman during one of his Caltech quantum physics lectures (which can be bought as audio files). It was his expression of hope that might soon be realised in the early '60s, that the huge intellectual and financial investment in understanding quantum physics, would yield some tangible value for mankind. So when Jon Butterworth responded to students' questions (and indeed as he pointed out periodically throughout his excellent talk) about the value in searching for the Higgs particle, I thought the idea of improving our knowledge is one noble aim, but how do we translate the enormous investment in time and money into practical returns? Clearly, as he said in his Q & A session, the knowledge gained from the LHC experiments are unlikely to materialise for some years, but what of the Science we have collectively supported as a species in the last 50 years?

The so called "impact" of research in all areas of endeavour has recently been added to the assessment "tick box" list for the UK Universities, in the form of the 2013 Research Assessment Framework. However, I am taking a different approach here and asking much simpler questions. For example, can we use our knowledge of physics and chemistry to predict the colour of a solution of a dye molecule? Can we predict with certainty, that the introduction of a sequence of synthetic nucleotides, derived from genomics research, lead to the successful and economically viable production of therapeutic molecules like insulin on an industrial scale? Can we predict the three dimensional structure of a simple protein molecule such as the enzyme ribonuclease, based on a knowledge of its sequence (or primary structure) in conjunction with the structural data bases of thousands of three dimensional structures of macromolecules,  the study of protein folding chemistry and physics and access to high level computing time (a la CERN?).In other words can we repeat the successes of teh Industrial Revolution in the way that say, steam power was harnessed? If we cannot, why and what, if anything, has gone wrong? Or are we (since we are all part of the wider Scientific community at the UTC) guilty of raising unreasonable expectations in respect of new treatments for diseases and new sources of renewable energy?

Copper sulphate and the dye Brilliant Blue, are difficult to tell apart when in a couple of unlabelled glass bottles sat side by side on a lab shelf. Clearly we can carry out a number of analytical tests to improve on the limited analytical qualities of our own eyes. We could for example, evaporate both solutions: large flat diamond-shape crystals will appear in one bottle, pointing at that is copper sulphate (so far a good result for science!). If, however, a textile company starting up next door asked if we could suggest, or even synthesise a molecule that had the same colour blue (let's forget its suitability for remaining "fast" on the dyed fabric through harsh washing conditions etc.) to enable them launch a new product. Could we advise them on the likelihood that they could perhaps even protect its chemical composition, possibly through a patent application? 

My first thought would be to look at a number of well known dyes, add one or two chemical substituents (theoretically at first, or as they say these days in silico) and calculate the impact they would have on the distribution of electrons between all of the atoms in the molecular structure. From this I would reach for the quantum mechanical equations that allow me to define the energy levels of all of the components, the bond lengths and angles etc. and then predict the outcome when a solution of the modified dye is dissolved in water. Not too much of a challenge then! In fact the theory behind this dates back nearly one hundred years, but the challenge of applying it to high molecular weight dyes requires significant computational power. I am not qualified to comment on the accuracy of the appropriate formalisms that are used to interrogate the theoretical structures, but I understand that the bigger the molecule the further it deviates from the observed (using methods like X-ray crystallography). In addition therefore, it is likely that absorption spectra in the ultra violet and visible region are unlikely to be predicted with a level of accuracy that guarantees we can "know" the colour of such in silico molecular constructs.

So I conclude that quantum theory gives us a good means of approximating the molecular design of a dye with a given colour. Thanks to the theoretical insights of Planck, Bohr, Heisenberg, Lehnnard-Jones, Hartree and Fock (as well as many others I am too ignorant to name) we have a framework on which to hang our experimental ideas, but I would say less of a predictive basis on which to synthesise some seemingly elementary molecules. As Hesisenberg reassures us, uncertainty remains! Still more work to be done there then? So how good is our model for gene expression and the subsequent translation of mRNA into a therapeutic protein in Escherichia coli? These issues are the result of much less theory and apparently simpler experiments. Nevertheless, the contributions of Jacob and Monod to developing the concept of the operon in genetics, together with the cracking of the genetic code by Brenner, Nirenberg and Khorana (and others again!), together with all that followed until the first genes were cloned in the 1970s, represents a formidable achievement of modern Science.

I vividly remember my first attempts to clone a gene and induce its expression in E.coli. It was in 1985 during a thoroughly enjoyable post-doc in Tom Bickle's lab at the Biozentrum in Basel. This was before mainstream PCR, so restriction enzymes, DNA ligases and a small pallet of expression vectors were available from friendly academic labs. The amazing work from the bacteriophage community had led to the  application of a small sequence of DNA duplex (a promoter) for driving expression of a gene placed upstream (ahead) of it, when the temperature of the culture was shifted from 30 to 42 degrees. I wont go into the details of how this works, but suffice to say, the experiment worked, the gene was expressed and the protein duly purified and characterised. So, the conclusion would be that our knowledge of the DNA sequences that provide for regulated expression of a protein in E.coli are sufficiently robust as to allow anyone to design a piece of DNA (part of our Synthetic Biology Toolbox?) that will lead to expression of a protein in E.coli. 

However, when I tried the same experiment on a second gene, weeks of repeated cultures proved fruitless. Plasmids were reconstructed, different experimental strategies were developed. No luck. The test of success: a large blue blob in the middle of an otherwise transparent slab of polyacrylamide, failed to materialise. I cant remember what happened in the case of the next gene and the one after, but success, or my hit rate, was possibly 25% in my favour. So, another partial success for Science, but not the comprehensive victory we demand from the theoretical base of Molecular Biology. Certainly not an ideal platform for launching  a programme of Insulin production, or a new venture in Synthetic Biology, you might suggest! Well, maybe its not that surprising and maybe as developing and practising Scientists, we should be more "honest". But more importantly we should consider our successes and failures as outcomes that require further analysis and outcomes that need to be communicated more precisely.

I am not going to give you more examples of the above, but I did mention earlier whether we have been able to harness our knowledge of structural biology and protein chemistry and physics to predict the structure of a protein from the gene sequence that encodes it. The answer is that we have had some successes, but many more failures. However along the way we are learning about the gaps in our knowledge and the goal is important if we are to design new drugs more efficiently to treat the looming health problems as the population expands and ages. I believe these endeavours are an essential part of the human condition, and that we have an intrinsic drive as a species to address the unknown. I also believe that communicating Science (and indeed all knowledge, literature, music and art) should be placed at the forefront of education. We must be able to challenge the science that we generate. I do not mean that we all need to be able to check through and validate the equations that form part of the experimental design of the Higgs experiments, but more are needed: maybe we should aim to  increase the number by 10 fold!

The time and effort taken by scientists like Jon Butterworth to meet with young students and in other fora, more senior citizens is so vital. Thanks to the fantastic work of the Liverpool Science Festival team and those of the senior staff at the UTC in Liverpool in bringing scientists of his calibre together with a whole array of outstanding Scientists representing a range of disciplines at all ages has been one of this year's remarkable achievements at the UTC.