Over the last ten years, there has been an exponential increase in the number of papers describing mutations in genes responsible for human disease. Clinicians are thrilled that this new information may translate into meaningful changes in health care for affected individuals. We carefully read such papers for clinical details to help determine which individuals and which families might benefit from molecular evaluation or testing. Sometimes these details are there, but more often they are not. Thus, clinical geneticists are frustrated by an inability to determine to which condition or conditions the reported gene mutation actually refers.

The project to sequence the human genome, nearing completion this year, has been a remarkable effort. It was powered in part by the promise that insights into human health and disease would result. The large number of excellent articles describing genes mutated in human diseases in this issue of Nature Genetics1,2,3,4,5,6,7,8 reflects the fruits (and hard work) that are to be celebrated. At the same time, however, they show how little we know about these genes, how much more work there is to be done and how urgently accurate and clear clinical descriptions are needed.

The clinical approach

A clinician can't help but wonder what was different about the 11 individuals with Ondine's curse (congenital central hypoventilation syndrome) in whom no mutation in PHOX2B was found, as described by Jeanne Amiel and colleagues1 on page 459, or what additional features might have been present in the 31 individuals with anophthalmia who had no mutations in SOX2, as outlined by Judy Fantes and colleagues3 on page 461. Did they have other structural or functional abnormalities? What was different about the pregnancies of those individuals? Has unusual behavior developed as they age? Was their mother's health or nutrition unusual before conception or during the pregnancy?

Often we read that the 'found' gene is actually one that was previously unknown. The gene structure may allow speculation about function, and there may be related genes that suggest clues as to pathway or pathogenesis, but the actual pathogenic mechanism must be determined through the hard work of defining time-specific, tissue-specific and species-specific expression and through building animal models of disease. The isoforms, protein folding, control mechanisms and position along a biochemical pathway await elucidation.

Important clues will probably come from full clinical descriptions of which tissues are involved, when the onset of disease occurs and what complications develop over time. Recognizing the signs and symptoms that identify exceptions should give hints about alternative pathways. For instance, as indicated by Catherine Dodé and colleagues2 on page 463, among the individuals with Kallmann syndrome who have mutations in FGFR1 (but not among those with other mutations), marked craniofacial structural anomalies are often seen. Those individuals with Smith–Magenis syndrome who have heart and renal defects are likely to have chromosomal deletions that include RAII, which is described by Rebecca Slager and colleagues8 on page 466.

Traditionally, clinicians approach their work quite differently from basic scientists. Medical school provides them with a broad background in anatomy, cell biology, physiology, pathology and therapeutics. When an affected individual or family presents with a problem, the clinician collects historical information, carries out a physical examination and gathers the results of laboratory and imaging studies. Then, because there may be hundreds of conditions to consider, the clinician usually consults the literature to arrive at a diagnosis that will allow appropriate counseling and therapy. By contrast, owing to training and background, the scientist selects a problem, formulates a hypothesis and then does experiments, collecting data to attempt to prove the hypothesis correct or incorrect. Although the two approaches have begun to overlap, because the clinician is dealing with human beings, there are many constraints and diagnosis and therapy are the goals.

The clinician needs clinical details and clues from the affected individual, the family and a literature review to determine which tests are appropriate to make a diagnosis. Gender, ethnic background, age and natural history help in that process. For example, for individuals with small heads (microcephaly), when should Seckel syndrome (see the paper by Mark O'Driscoll and colleagues5 on page 497) be considered? Historically, individuals with Seckel syndrome have been described as having intrauterine growth retardation (small at birth for their gestational age) and a head size that is much smaller than would be expected. There are many conditions that share these two features9. Thus, the description of additional clinical features, such as beaked nose, large-appearing eyes and sociable personality, may help the clinician to determine who should be tested for mutations in ATR, which O'Driscoll et al.5 have found to be associated with Seckel syndrome. These 'newfangled' DNA-based tests are expensive (if you can even get them done). Insurance companies don't like diagnostic 'fishing expeditions'. The clinicians tries to focus diagnostic testing and do only the most appropriate test(s) in a rational and justifiable manner. This requires knowing the indications and the clinical features that might implicate a mutation in a specific gene. If the clinical features reported have been unclear, the testing of appropriate individuals is unlikely to occur.

What information is needed?

“These are meant to be molecular papers,” you say. “Clinical information should be in a clinical journal.” “It takes too much space.” “We're not interested in those fuzzy details.” Without clear clinical descriptions, however, the relevance for human disease, diagnosis and therapy may be lost—and, after all, that is part of the purpose of such work. The clinician usually has a systematized way of collecting information (see box) and storing it for future reference (medical records). Some papers that report mutations in a human gene include tables giving phenotypic features2,3,7,8. But too few papers include the clinical features of those individuals in whom no mutation was found. But what is or was different about those individuals? Occasionally, DNA samples are obtained from cell repositories that are notorious for lacking good, verifiable clinical information—yet another problem! As genomic research moves towards proteomics, and proteomics moves towards 'proteotype–phenotype' correlations, clarity concerning the clinical features will become even more important.

Unfortunately, clinical descriptions are rarely static. Just as the individual changes with age, so do the clinical features. The study of the natural history of a genetic disorder is reflected against the normal maturational processes (physiologic changes that occur with aging) of human beings. How the gene product fits into various biochemical pathways will, in the long run, lead to the signs and symptoms of the disease as normal processes go awry. Thus, recording information about the natural history requires one to cast a wide net and often results in unexpected correlations. But that is the real biology, the real secret of gene function and the future challenge.

Now or never

All research on human gene mutations requires collaboration with clinicians. The clinicians are responsible for making the diagnosis, getting permission, obtaining the DNA samples and recording relevant clinical features. When papers describing mutations in genes responsible for human diseases are published, clinical details must be there. Experience tells us that if the full clinical descriptions have not been collected and recorded on the affected individuals whose DNA has led to identification of a mutation, they never will be (it is very difficult to contact people again). Authors need to insist that the clinical information collected by their colleagues be published or at the very least be readily accessible. Ideally, it should be part of every publication, although an accessible website is another option. If this information is not available, opportunities to advance our understanding of biology and human disease and to help affected individuals will be lost.