We have noticed, among some of our biological collaborators, a tendency to view the massively parallel methods of functional genomics as a highly efficient large-scale application of methods that they have already applied. For instance, gene expression profiling and polymerase chain reaction (PCR) are all methods that have been used by molecular biologists for decades. What we hope the reader will obtain from this book is an appreciation of how the near-comprehensive (and soon to be truly comprehensive) nature of the functional genomics approach as permitted by expression microarrays changes qualitatively and fundamentally the nature of biological investigation. Before our potential readers with biological backgrounds become offended by or disgruntled with this assertion, let us assure them that we present an equivalent critique for the purely computationally oriented bioinformatists and genomicists in the following section. Functional genomics is not, as some have portrayed it, a hypothesis-free fishing expedition, nor is it, even more charitably, only a hypothesis-generating enterprise requiring subsequent biological validation. It is fundamentally different in that it permits the posing of large questions that are grounded in an essential biological understanding of a particular domain. Unlike the questions posed in "traditional" genetics or molecular biology, these questions have less stringent requirements for prior supposition or claims of the role of a particular gene or metabolite in a biological process. An example of some of the broad questions that can be asked are:
• Which of all the known genes have regulatory mechanisms that appear to be similar to those regulated by the sonic hedgehog transcription factor in the cerebellum?
• Given the effect of 5000 drugs on various cancer cell lines, which gene singly is the most predictive of the responsiveness of the cell line to any chemotherapeutic agent?
• Given a known clinical distinction, such as that between acute lymphocytic leukemia and acute myelogenous leukemia, what is the minimal set of genes that can most reliably distinguish these two diseases?
• Is there a group of genes that can serve to distinguish the outcomes of patients with large B-cell lymphoma that are otherwise clinically indistinguishable on presentation?
• What distinguishes the signaling pathways of two of the substrates of the insulin receptor?
These questions are important biologically and clinically, and yet they can only be posed reasonably if they involve a comprehensive view of genomic regulation and involve the use of computational methods that can efficiently sift through the vast quantities of genomic data generated by the experiments required to answer these questions. Another way to consider functional genomics is to view it as serving as a filtered funnel through which these broad questions can be strained. The residue that remains is high-yield, detailed, and contains particular biological questions that are answerable by more traditional genetic or molecular biology techniques. This is illustrated in figure 1.6 below. The utility of this metaphor is as follows. The universe of possible participants in any given biological regulatory mechanism is finite but very large. Even with the most comprehensive in-depth expertise, a biologist may find be surprised about insights obtained through data mining of expression patterns, human genome sequences, and often from data obtained from other species. Without a genomic approach to guide her experiments, this biologist may expend several months in false leads or alternatively miss an important component to the system under study. Similarly, if the biologist is looking at a set of genes that are thought to be predictive of a given clinical condition, such as transplant rejection or cardiac disease, without the comprehensive view brought by functional genomics, elements of the diagnostic or prognostic procedure, such as the concentration of a gene transcript or a protein, may be omitted with a concomitant decrease in the sensitivity and specificity of the prognosis or diagnosis.
Figure 1.6: The functional genomics investigation as a funnel for traditional biological investigations. Broad questions and comprehensive data are the mix in which bioinformatics techniques are filtered to separate high-yield hypotheses or candidate genes from spurious findings and poor-quality hypotheses.
Was this article helpful?