As the proportion of older people in the global population increases, so too does the burden of chronic conditions such as cardiovascular disease, cancer, Alzheimer's and osteoarthritis. Biological markers that identify disease risk, flag occurrence before symptoms show, determine prognosis or guide treatment choices for individual patients would dramatically reduce this burden.

Biomarkers are not a new concept. It is well known that cholesterol levels indicate the risk of cardiovascular disease, that elevated levels of liver enzymes in the blood signal liver damage, and that a particular genetic variation predicts Huntington's disease.

But the past decade has seen a revolution in '-omics' technologies. High-throughput approaches (genome-wide association studies, transcriptomics, proteomics and metabolomics) coupled with bioinformatics are accelerating the pace of biomarker research. It is hoped that these studies will not only reveal new biological 'signatures' of disease to aid diagnosis and the implementation of personalized medicine, but will also shed light on the underlying mechanisms that trigger or perpetuate the condition.

Some successes have already been achieved, particularly with cancer. Tests such as MammaPrint and Oncotype DX, which measure the expression patterns of 70 or 21 genes, respectively, are now commercially available to help predict the risk of breast tumour recurrence and to guide treatment decisions for patients with breast cancer.

But most new biomarker candidates fall by the wayside and are not pursued — even those first published in high-profile journals. Identifying potential candidates is only the first step. The initial validation must be done to a sufficiently high standard and reported in enough detail for the study to be assessed and reproduced, thereby allowing the best candidates to be identified and taken further. Validation seems to be a major stumbling block.

As with all translational research, biomarker discovery involves two communities: basic scientists working to produce new technologies or to understand the cellular and molecular underpinnings of disease, and clinicians who seek to derive new insights from patient-oriented studies. Global recognition of the importance of translational research has resulted in renewed funding and infrastructural initiatives to pull these two communities together — the conventional boundaries between bench and bedside are becoming less distinct.

As more of Nature's readers and authors engage in translational research, we are encouraged to see an increase in submissions on potential disease biomarkers that address important clinical challenges. One of the main criticisms from our referees of such studies is that the methods are not presented in sufficient detail for the studies to be effectively refereed or repeated. We ask authors to ensure that, where patient studies reveal potential disease biomarkers, the initial validation is conducted to a robust standard and all methods and statistical analyses are reported in sufficient detail to allow the study to be repeated. Specifically, the manuscript should contain detailed information about the patient and control cohorts, the criteria for inclusion of patient samples and the methods for obtaining and preparing the samples. Any algorithms used to generate a signature should be described in sufficient detail to allow repetition.

These are exciting times in biomedical research, and Nature as a multidisciplinary journal is keen to have a role in bridging laboratory research with patient care. The technology exists to provide meaningful biological indicators of disease. If we are to harness this information to generate clinically useful diagnostic or prognostic tests to guide treatment decisions in the clinic, scientists straddling the boundary between bench and bedside must conduct and report their research with the rigour that each individual community expects.