Taken together, I was impressed with the ingenuity and methodology shown by Stiffler et al. in this paper. Actually, the approach they used is an excellent portrayal of the scientific method: their model generated a "hypothesis" (prediction) that was then empirically tested (fluoresence polarization) and subsequently refined, only to be repeated again and again to further resolve the question at hand. Such an approach strengthened not only their model but also my confidence in their approach to challenging a previous assumption (PDZ functional classes). This outcome also shows that the scientific method can function independently of the popular reductionistic approach to understanding biology, and the systems biology approach deserves more attention (in my opinion, at least). Furthermore, their work has generated novel data and a new interactomics platform (MDSM), both of which deserve replication and further experimentation by other labs. In addition, I felt the authors excercised appropriate discretion in selecting which figures to include in the main article and which to provide in the supplementary section; rarely did I wish to see a figure that was not found in the main text.
However, a few concerns did come to mind when I read this paper. First, the peptides they generated represented only the components of the proteins that may interact with the PDZ domains. What if the full protein interacts differently with the PDZ domain for reasons such as steric hindrance or alterations in folding? Furthermore, as reported in NCBI, some PDZ domains bind to other places than just the c-terminus of peptides, such as thepeptides' internal components, lipids, and even other PDZ domains. It is possible that such information was not reported until after the publication of this paper (in 2007), but regardless, more PDZ domain-peptide interactions may be present than MDSM can predict. That said, additional interactions would not destroy the predictive function of their model, as it is was aimed to and tested with the same specific peptide sequences. Consequently, this concern does not detract from their findings in my mind, and the researchers could potentially update their model if new types of interactions were characterized.
Overall, once I successfully waded through the dense material housed in this paper, I was impressed with the authors' explanations and found the paper an enjoyable read. And, on a more personal level, systems biology and proteomics papers no longer appear so daunting, either! All good outcomes indeed.
Email questions or comments to kaswart@davidson.edu
© Copyright 2011 Department of Biology, Davidson College, Davidson, NC 28035