January 2, 2006
Additional reading:
Cory Casanave's paper on Data
Access
work on ontology for biological signal pathways
EU’s program to model complexity using ontology
At:
There are two additional issues with recursive complex definition.
The first one is for complex comparison. A flat definition makes it quite
easy check for equivalence or compute the difference between two complexes. A
recursive definition is an order of magnitude more complex to handle. There
could be a huge number of equivalent combination. Also think how to handle this
kind of recursive definition in a relational database.
The second issue is about semantic. Complex organization is not complex
definition. If we add some information about complex organization, what about
defining a dedicated structure to handle it properly.
Chris.
(To Bio-PAX e-forum)
please define, for me, "flat definition".
Given that I understand the discussion, I would agree with Chris that a dedicated structure is needed in those cases where the object of study is the composition and aggregation processes leading to a molecular complex having a specific function; AND the flat definition is not sufficient.
I think I know the issue to be related to (sub)-structure to function issues.
In some cases the aggregation process is very well specified - and in this case the "flat definition" of the complex is a data encoding template with slots for both the components that are constituents of the complex as well as a specific and well defined sequence of events leading to the complex being completed. This type of structure to function aggregation can be handled with a flat definition?
In other cases, there is significant variation in the aggregation sequence as well as in the reactants that are involved leading up to the complex being created.
The two cases are "ontologically" different, the first one being simpler to understand. The data slots in your data classes are then merely places for metadata (such as a name) and for characterizations - such as concentration of a specific protein (for example, if this type of data ontology is to be used in micro-array data standardization).
In the second case the casual elements are inducing degenerate states (Edelman's term) which (likely) are instantiated in specific instances with substrate elements that just happen to be present in the environment.
But the overall "function" of the complex is held consistent even though "how" the function is achieved is not consistent from instance to instance.
QSAR (Qualitative Structure Activity Relationship) analysis needs to know if the complex is "simple", ie is well represented by a deterministic processes, or "complex" ie is not produced in the same way with the same ingredients each time. I have done some work on QSAR logics.
http://www.bcngroup.org/area3/pprueitt/kmbook/Chapter6.htm
This variability of the aggregation process is what my group is trying to study in the context of complexity as opposed to well defined and fully simple aggregation processes. I have studied and published in biologically feasible mathematical models of neural and immune processes, and so am aware of the general issues related to signal pathways.
I hope I am not creating any confusion.
Paul Prueitt (PhD in mathematics)