'Omics' data have increased very rapidly in quantity and resolution, and are increasingly recognized as very valuable experimental observations in the systematic study of biological phenomena. The increase in availability, complexity and nonexpert interest in such data requires the urgent development of accurate and efficient dimensionality reduction and visualization techniques. To illustrate this need for new approaches we extensively discuss current methodology in terms of the limitations encountered. We then illustrate a recent example of how combinations of existing techniques can be used to overcome some of the present limitations, and discuss possible future directions for research in this important field of study.