Sillycon valley fever

Over the past year, and especially over the past few months, there have appeared intimations that the newest revolution in post-genomic biology is under way. This is taking the form of microarrays of nucleic acid molecules, or DNA chips, onto which complementary probes can be annealed. Their main application is in the generation of a very large amount of data on gene expression. The activity in this field is growing rapidly and includes companies making chips or making chip makers and chip readers, and companies selling the DNA or reagents or software. I do not want to decry technical advances in biological research, nor do I want to retain old habits simply because they are old and I am becoming a sentimental old fool. But as one of the few voices from the pre-genomic era, and because the founders of any church know its defects better than all the new, enthusiastic converts, I am moved to say my piece. Like many new vogues, the new revolution has its epicentre in California, although some minor eruptions in the vicinity of Boston have been recorded. It is all accompanied by many proclamations in Old Sloganic: such as, “Why study one gene at a time when you can look at thousands of genes under dozens of different conditions at the same time.” There are new concepts such as ‘self-organizing data’ and ‘emergent phenomena’. The main idea, if it can be called that, is to take millions of data points and put them through some computer program, sometimes called ‘cluster analysis’, and see what association can be found. Some have even hinted at Fourier analysis of the data. But most iniquitously of all, one of the missionaries of the new field has stated that it will liberate us from the “shackles of hypothesis-dominated biological research.” In plainer words, you do not have to think anymore to do research. Are we really about to enter the decadent phase of biology in which scientists will be unable to see what the problems are or, if they do, will be unable to formulate questions that could be answered, either by observation and measurement, or by intervention and experiment? It is my view that you cannot study gene expression and make sense of the results without having an explicit theory about global aspects of gene regulation. We are told that when we study all genes we see many unexpected changes in expression, even of genes we thought to be the very epitome of constancy. Taking all these changes seriously, however, implies the hidden assumption that they have been fixed by natural selection and are optimal for that organism. If this were so, we can calculate from the data the evolutionary cost in control genes required to specify all of the combinations found. This is enormous. Let me briefly sketch an alternative approach. Before we consider how genes are turned on and off, we first have to ask how the transcription resource is allocated to all the genes active under some given circumstance. Clearly, as specified in its DNA sequence, each promoter will have some affinity for the transcription complex. This can be adjusted by natural selection under standard conditions of growth so that the majority of genes making small amounts of message can access the polymerase in competition with the smaller number of genes producing abundant messages. Now, if a change in the population of messages occurs as a result of some change in the set of genes expressed, the absolute amount of RNA produced by any gene will change in response to the new competitive conditions set up. The changes do not reflect a new set of specifications for the genes, but arise automatically as responses to a global situation. We can take this further and think of competition for RNA processing, for access to ribosomes, and so on. Thus I do not doubt the significance of all of these findings from the point of view of measurement, but I doubt their meaning for the biology of the cell, and whether they can be comprehended without a theory of the molecular ecology of the cell. We may also find genes which, although turned on in a cell, are unnecessary for the functioning of that cell simply because they obey a ‘don’t care’ condition for that cell. Nature simply may not bother to turn them off, even though they are not needed, as long as they are doing no harm. For survivors of the pre-genomic era, this approach will be very familiar: I now have the outlines of a theory that I can develop further. Moreover, I can construct critical tests for its consequences. Running more chips and having more data and more computer programs will not extract this theory because computers can’t think. Sadly, human programmers are becoming like their machines. Perhaps what I can do with my approach is to find the critical genes to study and give the world a smaller chip with only a handful of genes. When next you hear “More is better”, just remember that Uncle Syd says “The least is best.” Let’s get back to solving problems and providing answers to questions. R671