Subset substances - EGFR example

  In the following examples, we use the following prefixes for name
spaces: bp for BioPAX Level 2, sbpax for SBPAX, rdf for the RDF
namespace. Objects without namespace are specific to this example.

  Consider :EGFR, a protein, as it is typically defined in a BioPAX file:

  :EGFR   rdf:type   bp:protein

  It is assumed that :EGFR refers to the unphosphorylated form, since
every reference to a phoshorylated form needs to include a reference
to a sequence feature, such as :pY1148-feature:

  :pY1148-feature   rdf:type   bp:sequenceFeature

  The complex EGF-EGFR-pY1148, which has the component EGFR-pY1148,
which is phosphorylated at site Y1148, reads in BioPAX:

  :EGF-EGFR-pY1148   rdf:type   bp:complex
  :EGF-EGFR-pY1148   bp:COMPONENTS   :component1
  :component1   rdf:type   bp:sequenceParticipant
  :component1   bp:SEQUENCE-FEATURE-LIST   :pY1148-feature
  :component1   bp:PHYSICAL-ENTITY

  EGFR is phosphorylated up to five times, which would require five
sequence features. Mentioning them with every reference is
impractical. Therefore, SBPAX allows expressing different phosphoforms
as substances. For example the singly phosphorylated EGFR-pY1148 would
be a protein and have the sequence feature :pY1148-feature:

  :EGFR-pY1148   rdf:type   bp:protein
  :EGFR-pY1148   sbpax:hasSequenceFeature   :pY1148-feature

  This looks similar to BioPAX Level 3, but there is a difference: In
BioPAX Level 2 and 3, if a feature is not mentioned, it is not clear
whether it is not present or whether its presence is unspecified. In
SBPAX however, features not mentioned are clearly features whose
presence is unspecified, which is the only way to avoid wrong
assertions on features not discovered yet. To specify that a feature
is not present, we would say

  :EGFR   sbpax:lacksSequenceFeature   :pY1148-feature

  Further, we can specify the union of :EGFR and :EGFR-pY1148, where
the presence of the feature is unspecified. This is important, since
in some interactions, both forms can participate. Calling this
substance :EGFR-aY1148, (a for "all") we can assert the other two to
be subset substances:

  :EGFR   sbpax:subSetOf   :EGFR-aY1148
  :EGFR-pY1148   sbpax:subSetOf   :EGFR-aY1148

  The substance :EGFR-aY1148 looks similar to an entity reference in
BioPAX Level 3, but the difference is that it can be used just like
other substances. Most importantly, there is a natural way to
integrate partially specified phosphoforms, if there are multiple
sequence features, of which EGFR can have up to five. Consider two
sites of EGFR, Y1148 and Y1173, and the four possible states
distinguished by these sites: EGFR (not phosphorylated), EGFR-pY1148
(only at site Y1148), EGFR-pY1173 (only at site Y1173) and
EGFR-pY1148-pY1173 (both sites). Each of these four forms is a
substance in SBPAX. EGFR and EGFR-pY1148 are subset substances of
their union EGFR-aY1148, which is in turn a subset substance of the
union of all four, EGFR-aY1148-aY1173. So there is a hierarchy of
subsets with three levels. With more than two site, the subset
hierarchy will have even more levels. Any of these may be needed as
part of a pathway, and SBPAX provides a natural way to organize them
all.

  Not only proteins have phosphoforms, but also complexes:

  :EGF-EGFR   rdf:type   bp:complex
  :EGF-EGFR-pY1148   rdf:type   bp:complex
  :EGF-EGFR-aY1148   rdf:type   bp:complex

  With simple references to component substances, specifying
components becomes a lot easier than in BioPAX:

  :EGF-EGFR   sbpax:hasComponent   :EGFR
  :EGF-EGFR-pY1148   sbpax:hasComponent   :EGFR-pY1148
  :EGF-EGFR-aY1148   sbpax:hasComponent   :EGFR-aY1148

  Most importanly, subset relationships between components neatly
propagate to complexes:

  :EGF-EGFR   sbpax:subSetOf   :EGF-EGFR-aY1148
  :EGF-EGFR-pY1148   sbpax:subSetOf   :EGF-EGFR-aY1148