Transcript The 2 flavours of part relationships*:

Representing Part Relationships
Between Developing Structures
Anatomy Ontologies – a MOD
prespective.
What literature curators need
1.The ability to query the ontology to home in
on candidate terms based on limited
available information.
Useful queries for this:

give me a list of all the types of X that are part of Y.

find where structures referred to by candidate terms are
located.
2. A way to curate even when the available
anatomical data is vague.
What database users need
 The ability to precisely extract biologically relevant
information from an ontology – rather than navigating
some convoluted DAG

E.g.- for any term X
 Locate X,
 what is X,
 what subtypes does X have
 What parts does X have
 Curations grouped accurately according to type and
part relationships.
Argument for making all part
relationships into integral_part
Cardinality
Anatomy Ontology terms can be classed according to the number
of structures per whole organism (C).
1. Many per org (C>1) – bristle, scale or neuron
- always possibility of further subdivision – e.g.neuron
% motor neuron
% ventral tp motor neuron
2. Fixed/known number per org: e.g.- limbs or (perhaps) segments
(C>1)
3. One per organism (C=1) – adult head
4. Less than one per organism - sexually dimorphic structures
(C=0.5)
The 2 flavours of part relationships*:
 X part_of Y: All instances of X are part of some
instance of Y. (symbol: <)
 Obligatory (?)
 Y has_part X: All instances of Y have some
instance(s) of X as a part. (symbol: >)
 When both of these conditions are satisfied, the
relationship is known as integral_part. (symbol: <>)
* For the sake of simplicity, these definitions avoid time/stage. These will
be dealt with later.
Example: sex comb only on male
prothoracic leg; all legs have claws
If we were only using part_of (<) then this is legal:
leg
% male prothoracic leg
< sex comb
< claw
With integral_part (<>), we are restricted to this:
leg
% male prothoracic leg
<> sex comb
<> claw
(% = is_a)
Deductions:
All legs have a claw as a part. Prothoracic leg is_a leg.  Prothoracic legs
have a claw
Sex comb part_of leg
How literature curation works
Curation of expression or phenotype with any
term X can mean:
expressed/having phenotype in all types of X
OR
expressed/having phenotype in some
unspecified subset of X (X is the most precise
term we can curate to, given the evidence presented
in the paper being curated)
Grouping – the need for has_part
FOR:


Gene 1 - expressed in X (subset)
Gene 2 - expressed in Y
 If the only known part relationship between X and Y
is:

Y part_of X


It is not safe to group these two curations - we don't know
whether the curation to X was made because of expression in a
type of X that has a Y as a part.
X has_part Y

Then these two curations can be safely grouped - all types of X
have a Y as a part.
Representing sexual dimorphism
 organism









<> gonad
% testis
% ovary
% male organism
<> testis
% female organism
<> ovary
<> head
<>brain
Part relationships during development
Types of Developing Structure
anlage
Contiguous tissue defined by lineage labelling as contributing all or the majority
of its cells to some specified mature structure but not (yet) having distinct
morphological boundaries.
primordium
Contiguous tissue defined by lineage labelling as contributing all of its cells to
one or a few specified mature structures and having morphologically distinct
boundaries
germ layer
Primary division of embryo established just prior to &/or during gastrulation.
Initially constituting a contiguous tissue contributing all of its cells a large but
limited set of mature structures.
compartment
Contiguous tissue defined by lineage labelling as consisting of cells unable to
cross a *compartment boundary* to mix with cells in a neighbouring tissue with
which it is contiguous during development.
Types of Developing Structure
These terms group multiple primordia and anlage over the
complete development of the system, part or organ.
Developing system (e.g.- developing nervous system)
Developing (cardinal) body part – e.g.- developing head
Developing organ (e.g.- developing brain)
Relationships linking stage to
anatomy
 Ts= stage n: starts during or after stage n
 Te= stage n: ends during or before stage n
e.g. central brain primordium; ts=6 te=8
Note: These definitions allow for cases where the transition between subtypes of a term occur spread out over multiple stages.
Identity and development

We need some concept of identity for continuants
(biological structures existing over time) that can
account for changes composition over time (X can
have diff parts at diff stages).

Ideally, shifts in identity during development, e.g.neuroblast -> neuron, will be based on intrinsic
criteria. This could be morphological (e.g.- having
an axon), or perhaps (?) functional (heart starts
pumping).
Intrinsic identity and part_of
developing nervous system ; ts=4 te=16
< neuron X ; ts=13 te=16
~ neuron X’ ts=16 …
< neuron Y ts=14 te=16
~ larval nervous system ts=16 …
< neuron X’ ts=16 …
Reasoning:
Part_of : all neuron X are part_of (some) developing nervous system
 - all neuron X after the stage that developing nervous system is
considered mature have to get a new name
- identity is being ascribed extrinsically.
Defining has_part for developing
structures
This definition cannot be used for part
relationships between a developing stucture
and parts it instantiates at different stages:
For Y has_part X: All instances of Y at all times
(stages), have some instance(s) of X as a
part
Part relationships during development
developing central nervous system; ts=3 te=16
<> developing brain; ts=3 te=16
<> central brain anlage; ts=5 te=5
~ central brain primordium; ts=6 te=8
<> central brain primordium; ts=6 te=8
Is there some definition of has_part that is still useful for grouping
curations and for reasoning, but that can be used here?
Possible alternative def for has_part
 For Y has_part X: All instances of X have some
instance of Y as a part during the stages that Y
exists.
e.g. - <> developing brain; ts=3 te=16
<> central brain anlage; ts=5 te=5
~ central brain primordium; ts=6 te=8
<> central brain primordium; ts=6 te=8
Tells us that developing brain has central brain anlage
as a part during stage 5 and central brain primordium
as a part during stages 6-8
 ie- given the stage, we can list reliably list parts
Poss solution to all – seemed like a
good idea in the pub last night.
Term X ts=4 te=8
<> Term Y ts=6 te =10
Y part_of X during the stages that both X and Y
exist: In this case stages 6-8
X has_part Y during the stages that both X and
Y exist: In this case stages 6-8
develops_from
As long as part of the definition of ‘B
develops_from A’ includes:
A and B abut in time: there is no instance that
is both A and B simultaneously.
Then we can use overlap between stages to
specify a range during which a transition
occurs. Or, if A(Te) and B(Ts) are adjacent
stages – we can tie the transition to a stage
boundary.
Tying develops_from transition to
stage
term X te=n
~ term Y ts=n+1
(note ts>n+1 would not be legal)
Implies that the transition from X to Y occurs at the stage
transition.
term X te=n
~ term Y ts <=n
Implies that the transition from X to Y occurs at some point
during the overlap in stages between X and Y.
Using cardinality in part reasoning
developing central nervous system C=1 ts=3 te=16
< developing brain C=1 ts=3 te=16
< central brain anlage ts=5 te=5 C=1
~ central brain primordium ts=6 te=8 C=1
< central brain primordium ts=6 te=8 C=1
All central brain anlage are part_of some developing
brain. But in any one organism there is only 1 central
brain anlage and one developing brain. So
developing brain must have central brain anlage as a
part during the stages that central brain anlage
exists.