Kitano et al. respond:

The first issue raised by Blinov et al. suggests that pathway maps are too simplistic to represent the protein combinatorial explosion in signal cascades. They detail Figure 3e in our article to illustrate their point; however, this figure was used solely to demonstrate the look-and-feel of how to represent pathways as process diagrams. Therefore, we used part of the diagram in a Hanahan and Weinberg paper1, which is also a pathway extensively used in simulation studies2,3. It was not argued that this was a comprehensive representation of the EGFR pathway. Our recent interaction map published in Molecular Systems Biology4 was intended to be a comprehensive EFGR map of experimentally validated interactions. We did not enumerate all possible interactions and molecular states and recognize that there are interactions not listed in the map due to lack of experimental validation, despite theoretical and intuitive possibilities. The process diagram is neutral on what should be described in the map. It defines the graphical representation of an interaction map; thus, the oversimplification critique does not apply to the process diagram itself as construction of these maps relies on experimental evidence.

The second issue raised was that describing all combinatorial states of molecules and resulting complexes would result in a combinatorial explosion making a rule-based approach more appropriate for modeling. We would argue that this depends on the intended use of the map. The process diagram was motivated by an experimentalist's need partly to represent detailed interactions, including residue modification state, to improve experimental design, and partly to visualize their data in the context of a pathway map where each combinatorial state has been explicitly described, regardless of the level of complexity. It is imperative that software tools make such complex and large-scale maps accessible to users.

Although the rule-based approach has attracted much attention as a viable approach for dynamical simulation5,6, it may not allow users to project experimental data on to each combinatorial state without expansion. As illustrated by Blinov et al. wherever the rule-based approach is shown to be effective, the process diagram can then be used to expand graphical notation to represent rules and the network generated from the rule. We would like to incorporate such features into the process diagram and are receptive to constructive critiques to create standard graphical notations; to this end, we have formed an international alliance to standardize graphical notation called Systems Biology Graphical Notation (http://www.sbgn.org/).