About BlackBox Pathways

Here is some use case as provided by Sujatha & Malabika Sarkar


E2 (estrogen) induced phosphorylation of STAT5A, STAT5B and STAT3


The catch is there is an unknown pathway which ultimately activates phosphorylation of both Stat3 and Stat5. One way to show this in BioPAX is to use control class to show an indirect control. However if there are multiple activators of this pathway ( and indeed there are although not listed in the above abstracts) it can need n*m control relations which is a bit of a mess. Besided it might have been nice to have a blackbox estrogen pathway as a placeholder for future curation.

Thus, I would like to express it in this form:

17B-Estradiol -<controls>-> Estrogen induced pathway -<controls>->Phosphorylation of Stat 3 and 5

It is almost possible to represent this in BioPAX, however second control is problemmatic. Since controller property requires a PEP

Now this is meaningful in level 2, since you need a PEP to capture state information ( location and modifications ). However stochiometry for a controller does not really make sense. This was extensively discussed in the SBML list as well. So for level 3 I propose to change the domain of the controller property to state or physical entity or pathway, which also solves the above problem.

last edited 2006-01-08 22:16:35 by Emek Demir