Alison Wright

Solar System: Captain Cook and the black drop.

resolved X-ray diffraction to a chemical reaction was the study of the structural changes that are associated with the photodissociation of the CO group in myoglobin, traced at the nanosecond timescale. Plech et al. have extended the diffraction technique to reach the picosecond timescale and track the photodissociation of iodine molecules dissolved in carbon tetrachloride. This reaction has been examined extensively as a model of the competition between two modes of recombination for the iodine atoms: whether the photodissociated atoms are trapped by the solvent and recombine with each other (called geminate recombination);or whether the iodine atoms escape their cage of solvent and recombine with other iodine atoms (non-geminate recombination). Plech et al. used 150-fs laser pulses to excite the iodine solute,generating a mixture of electronic states that mainly dissociate to ground-state atoms. The excited iodine molecules that are formed have a bond length of 0.27 nm, which stretches to 0.4 nm in less than a picosecond. Next, the molecule either dissociates into two iodine atoms or loses energy to the solvent, relaxing to an excited bound state on a timescale of 2.7 ns or to the ground state in about 180 ps. By sending intense 100-ps pulses of X-rays into the iodine solution at varying time intervals after the laser excitation pulse, Plech et al. mapped changes in the bond lengths between iodine atoms and between neighbouring solvent molecules during the course of the reaction, through the series of diffraction patterns created. On the question of geminate recombination, Plech et al. find that for only 14% of excited iodine molecules do the dissociated atoms escape the solvent cage to recombine with an atom other than their original partner. But a more striking feature of their results is the signature of structural variation in the solvent during the reaction (Fig. 1). When the reaction is probed on a spatial scale of around 0.6 nm, the solvent structure is seen to change slowly for about 20 ns, and then to change rapidly over a period of about 50 ns. But when the reaction is probed on a smaller spatial scale of 0.15 nm, the solvent structure changes at about the same rate for several hundred nanoseconds. The news and views

Fluid mechanics: Impact factors

To design a fire-extinguishing system, it’s useful to know how water droplets behave when they collide at speed with hot surfaces. A new study reveals shortcomings in the present theoretical description.

Dance: Einstein in motion

741 cause, and mere necessary machinery, that so much of the barren disputes of biology are due,” wrote Huxley. It would seem that, with such a sloppy and essentially misguided interpretation, these disputes would not be solved in a hurry. Unfortunately, even Tinbergen’s careful analysis of cause and function could not prevent a confusion of concepts that continues to this day. Some will say that ethology is no longer a scientific discipline in its own right, but that depends on who you ask: a behavioural ecologist and a cognitive ethologist might give you different answers. Nevertheless, Burkhardt notes that the core ideas of classical ethology dissipated astonishingly rapidly; few contemporary ethologists would use such concepts as ‘action-specific energy’, for example. This discarding of outmoded ideas would seem natural for any vibrant scientific discipline. Burkhardt rightly maintains that it was the empirical and theoretical approach introduced by Lorenz, Tinbergen and their colleagues that made the study of animal behaviour what it is today. In order to study the genomic or neural mechanisms of behaviour, we need to know how behaviour works, and for that an ethological analysis is crucial. This wonderful book shows very clearly how early ethologists made such analysis possible. ■

Planetary science: Stardust's comet memories.

NATURE | VOL 428 | 25 MARCH 2004 | www.nature.com/nature 381 knowledge obtained from ‘omics’techniques into clinical application, researchers are attempting to use microarrays — a means of analysing patterns of gene expression by looking at the mRNAs present — to determine the appropriate treatment for different patients. For instance, microarray profiling of breast tumours, using a set of 70 informative genes, has revealed gene-expression signatures that are associated with a good or a poor prognosis. In these studies, roughly 40% of early-stage breast cancers turned out to have a ‘good’ signature, associated with only a 15% risk of metastasis (the spread of tumour cells) and a 5% risk of death by 10 years after diagnosis. According to current clinical practice, most patients with earlystage breast cancer receive adjuvant therapy (chemotherapy or endocrine therapy after surgery). But less than 1% of the patients with a ‘good prognosis’ signature would be likely to benefit, and the treatments often have harmful side effects (discussed by L. Van’t Veer, Netherlands Cancer Institute, Amsterdam). So researchers now suggest using the ‘poor’ signature to guide the administration of adjuvant therapy to those for whom it would be most useful. Indeed, gene-expression profiling is already being introduced clinically as a diagnostic procedure in academic centres in the Netherlands (L. Van’t Veer), with similar programmes being launched in the United States.The profiles have been validated retrospectively in several cohorts of patients. Prospective validation, however, can only be achieved after 5–10 years. Also — as several speakers discussed — some technical issues have yet to be resolved, including how to procure and process samples and to ensure reproducibility and quality control. Nevertheless, various analysis platforms and short lists of diagnostic genes are being developed to predict the prognosis associated with other types of tumours, such as lymphomas (L. Staudt, NCI, Bethesda, and M. Piris, CNIO, Madrid), or to predict the likelihood of particular endpoints, such as metastasis (T. Golub, Broad Institute, Cambridge, Massachusetts). Predicting a patient’s response to specific types of treatment would, of course, be the most important application.But it may take years for the approach to meet with regulatory approval and be extended from academic centres to mainstream practice. Another strand of research involves the proteomic profiling of blood serum. This could soon complement — if not replace — the use of established tumour ‘markers’ in cancer screening and early diagnosis (E. Petricoin, FDA, Bethesda). Whereas specific tumour markers are often used singly, proteomic serum profiling involves the use of mass spectrometry to identify up to 15,000 peaks, representing proteins and protein fragments that are defined by their mass-tocharge ratios. The sensitivity and specificity of such profiles has exceeded 90% for the diagnosis of lung cancer, and approached 100% for ovarian cancer (E.Petricoin).Most of the peaks have still to be identified, and several investigators argued that panels of specific immunoassays, which use antibodies that recognize particular proteins, should be developed for diagnosis instead. But it is the fragments, not intact proteins, that are often the most informative diagnostically, and there may be dozens — if not hundreds — of such fragments, possibly existing in complexes with one another and with other serum proteins. That makes the development of specific assays demanding. A further application of proteomics concerns the analysis of signalling pathways, for example by studying the levels of specific phosphorylated proteins — a key feature of many pathways — in tumour samples (E. Petricoin). Analysis of these signalling profiles is being incorporated into many clinical trials to identify biological markers that indicate which tumours are likely to respond to treatment. Proteomic tumour profiling is often carried out from specific cell types, acquired by microdissection of the tumours. Such studies could also help us to understand the tumour microenvironment, where many different cell types and interactions may affect tumour behaviour. For example, in breast cancer, characteristics of the stromal fibroblast cells adjacent to the tumour might change, and contribute to, cancer progression (R.Weinberg,MIT) — much as does the growth of new blood vessels. Moving on to studies of gene sequence (genomics), mutations in roughly 1% of human genes (291 genes in total) were reported to contribute to cancer (M. Stratton, Sanger Centre, Hinxton). Twentyseven of these genes, more than would be expected by chance, encode protein kinases — enzymes that phosphorylate other pro-

Art: Imagine space

The NASA Art Program has launched ‘Copernica’, a collection of specially commissioned artwork reflecting the wonder of the human adventure in space. But this online gallery is a little different: you can explore the collection interactively by navigating through a universe interpreted by artists.

Solar System: Close encounter of the cometary kind

A sequence of images from the Solar and Heliospheric Observatory has captured the approach of a comet to within 15 million kilometres of the Sun.

Vote for Little higgs

The Higgs boson has eluded discovery, so are physicists starting to doubt it exists? A reworking of ideas has produced the ‘Little Higgs’ theory, an attractive solution that suggests a discovery is almost within reach.