From e33a31bd5ae61df0f609c12023cb43864a74de21 Mon Sep 17 00:00:00 2001 From: David Osumi-Sutherland Date: Sat, 26 Sep 2020 10:56:33 +0100 Subject: [PATCH 1/5] Create relations_guide.md Unfinished --- documentation/relations_guide.md | 85 ++++++++++++++++++++++++++++++++ 1 file changed, 85 insertions(+) create mode 100644 documentation/relations_guide.md diff --git a/documentation/relations_guide.md b/documentation/relations_guide.md new file mode 100644 index 000000000..185738a97 --- /dev/null +++ b/documentation/relations_guide.md @@ -0,0 +1,85 @@ +## A guide to what relations to use where in the Cell Ontology + +The cell ontology is not a database for recording all properties of cell types. We can't record all genes expressed in a cell type, or all synaptic connectoins. Some judgement is therefore necessary when adding relationship in the cell ontology - are they necessary to define a cell type? + +Relations have their own heirarchy, with more general relations at the top and more specific ones underneath. If you use a specific relation, then the + +### Recording location + +We record anatomical location by linking to terms from Uberon. + +For most purposes we record the anatomical location of cells using **'part of'**. + +e.g. eipthelial cell 'part of' some eipthelium + +means that +(a) All epithelial cells are part of an epithelium. We wouldn't say 'epithelial cell' part_of some 'kidney tubule epithelium', because not all of them are. +(b) All parts of an epithelial cell are part of an epithelium. +(c) Epithelial cells are part_of some epithelium at all times. This last stricture can be hard to apply in the context of development. Some judgment may be required, e.g. - (TBA) + +Some cells, most obviously neurons, often do not have all parts in an anatomical structure we want to relate them to. We have a general relation for this: **'overlap'**, but often we want to say something more specific. For example, neuon types are often referred to in part by the location of their soma, for example anterior horn cells have a soma in the anterior (ventral) horn of the spine. We have a dedicated relation for this: 'has soma location'. + +We also have a dedicated set of relations for recording the location of synaptic terminals and projections of neurons. See [Relations for neurons](#Relations_for_neurons) for details. + +### Taxon constraints + +in_taxon + +e.g. + +The following should be recorded as annotation_property axioms as they are shortcuts for more this allows are pipelines to safely use them in reasoning. + +only_in_taxon # Not currently true. Obviously cannot be used in nested class expressions if this is true. + +never_in_taxon + +Further reading: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC3098089/ + +### Recording function + +We record cellular function by linking to GO biological process terms using the relation (objectProperty) **'capable of'** + +### Recording developmental lineage + +We record developmental lineage relationships between cell types using **develops from**, or where we are sure there are not intermediates between the related cells, by using **'develops directly from'** + +### Relations for neurons + +#### Synaptic connectivity + +To record neuron to neuron or motor neuron -> target cell connectivity use. Use these relations sparingly where connectivity is key to definition, e.g. motor neuron types defined by the type of muscle cell they synapse to. + +**synapsed to** - preferred direction to record, as it fits with + +**synapsed_by** - Useful in cases where all X synapsed by some Y but there reverse is not true + +To record connection between a neuron and a region it innervates we have a number of relations, all sub properties of overlaps + +![image](https://user-images.githubusercontent.com/112839/94337631-e0a83300-ffe3-11ea-8f13-ac8a484a5fb3.png) + +Historically we have used *has synaptic terminal in* + + +### Recording cell markers + +has part + . has plasma membrane part + +expresses + . has plasma membrane part + + Absence of a marker + lacks_plasma_membrane_part + +#### A note on cell markers and taxon constraints + +If you choose a cell specific marker + +#### A note on when to record cell markers + +# Recording cell shape or other morphological qualities + +'bearer of' some + + + From 7beed79553682c1c531a37ffe48541cee05687c2 Mon Sep 17 00:00:00 2001 From: David Osumi-Sutherland Date: Sat, 26 Sep 2020 18:05:46 +0100 Subject: [PATCH 2/5] Update relations_guide.md --- documentation/relations_guide.md | 28 +++++++++++++++++++++++----- 1 file changed, 23 insertions(+), 5 deletions(-) diff --git a/documentation/relations_guide.md b/documentation/relations_guide.md index 185738a97..eb631ff78 100644 --- a/documentation/relations_guide.md +++ b/documentation/relations_guide.md @@ -23,22 +23,36 @@ We also have a dedicated set of relations for recording the location of synaptic ### Taxon constraints -in_taxon +in_taxon some +only_in_taxon some e.g. +TBA -The following should be recorded as annotation_property axioms as they are shortcuts for more this allows are pipelines to safely use them in reasoning. +The following should be recorded using an annotation_property axiom: -only_in_taxon # Not currently true. Obviously cannot be used in nested class expressions if this is true. +never_in_taxon -never_in_taxon +e.g. + +TBA Further reading: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC3098089/ +Inference of taxon constraints from other ontologies: + +Relationships to taxon constrained terms from other ontologies can result in inferred taxon constraints. + +Examples: + - Inference of Taxon constraints from taxon constrainted makers: + - Inference of Taxon constraints from taxon constrainted anatomy: + ### Recording function We record cellular function by linking to GO biological process terms using the relation (objectProperty) **'capable of'** +e.g. 'capable of' some 'androgen secretion' + ### Recording developmental lineage We record developmental lineage relationships between cell types using **develops from**, or where we are sure there are not intermediates between the related cells, by using **'develops directly from'** @@ -79,7 +93,11 @@ If you choose a cell specific marker # Recording cell shape or other morphological qualities -'bearer of' some +e.g. erythrocyte bearer_of some biconcave + +# Recording cellular qualities (eg. ploidy, nuclear number) + +e.g. 'enucleate erythrocyte' EquivalentTo erythrocyte and ('bearer of' some anucleate) From a7aa38cace32bccd4a563dd045fa392522844c03 Mon Sep 17 00:00:00 2001 From: David Osumi-Sutherland Date: Sat, 26 Sep 2020 18:06:09 +0100 Subject: [PATCH 3/5] Update relations_guide.md --- documentation/relations_guide.md | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) diff --git a/documentation/relations_guide.md b/documentation/relations_guide.md index eb631ff78..a0ee4c9d7 100644 --- a/documentation/relations_guide.md +++ b/documentation/relations_guide.md @@ -91,11 +91,11 @@ If you choose a cell specific marker #### A note on when to record cell markers -# Recording cell shape or other morphological qualities +### Recording cell shape or other morphological qualities e.g. erythrocyte bearer_of some biconcave -# Recording cellular qualities (eg. ploidy, nuclear number) +### Recording cellular qualities (eg. ploidy, nuclear number) e.g. 'enucleate erythrocyte' EquivalentTo erythrocyte and ('bearer of' some anucleate) From 6fe7e10f80da17b5ff411ee06db395fcc77b29b4 Mon Sep 17 00:00:00 2001 From: David Osumi-Sutherland Date: Thu, 1 Oct 2020 14:48:44 +0100 Subject: [PATCH 4/5] Update relations_guide.md Incremental improvements --- documentation/relations_guide.md | 49 +++++++++++++------------------- 1 file changed, 20 insertions(+), 29 deletions(-) diff --git a/documentation/relations_guide.md b/documentation/relations_guide.md index a0ee4c9d7..f91f5271a 100644 --- a/documentation/relations_guide.md +++ b/documentation/relations_guide.md @@ -1,62 +1,53 @@ -## A guide to what relations to use where in the Cell Ontology +# A guide to what relations to use where in the Cell Ontology -The cell ontology is not a database for recording all properties of cell types. We can't record all genes expressed in a cell type, or all synaptic connectoins. Some judgement is therefore necessary when adding relationship in the cell ontology - are they necessary to define a cell type? - -Relations have their own heirarchy, with more general relations at the top and more specific ones underneath. If you use a specific relation, then the - -### Recording location +## Recording location We record anatomical location by linking to terms from Uberon. For most purposes we record the anatomical location of cells using **'part of'**. -e.g. eipthelial cell 'part of' some eipthelium +e.g. eipthelial cell SubClassOf **'part of'** *some* eipthelium means that -(a) All epithelial cells are part of an epithelium. We wouldn't say 'epithelial cell' part_of some 'kidney tubule epithelium', because not all of them are. -(b) All parts of an epithelial cell are part of an epithelium. -(c) Epithelial cells are part_of some epithelium at all times. This last stricture can be hard to apply in the context of development. Some judgment may be required, e.g. - (TBA) -Some cells, most obviously neurons, often do not have all parts in an anatomical structure we want to relate them to. We have a general relation for this: **'overlap'**, but often we want to say something more specific. For example, neuon types are often referred to in part by the location of their soma, for example anterior horn cells have a soma in the anterior (ventral) horn of the spine. We have a dedicated relation for this: 'has soma location'. + 1. All epithelial cells are part of an epithelium. We wouldn't say 'epithelial cell' part_of some 'kidney tubule epithelium', because not all of them are. + 1. All parts of an epithelial cell are part of an epithelium. + 1. Epithelial cells are part_of some epithelium at all times. This last stricture can be hard to apply in the context of development. Some judgment may be required, e.g. - (TBA) -We also have a dedicated set of relations for recording the location of synaptic terminals and projections of neurons. See [Relations for neurons](#Relations_for_neurons) for details. +Some cells, most obviously neurons, only have some parts in the anatomical structure we want to relate them to. For example, anteriior horn motor neurons have a soma in the anterior (ventral) horn of the spine, but also project out of the spine to innervate muscles. We have a general relation for this, **'overlap'** (has some part in), but often we want to say something more specific. For example, neuon types are often referred to in part by the location of their soma. We have a dedicated relation for this: **'has soma location'**, allowing us to record: -### Taxon constraints +'anterior horn motor neuron' SubClassOf **'has soma location'** *some* 'ventral horn of spinal cord' -in_taxon some -only_in_taxon some +We also have a dedicated set of relations for recording the location of synaptic terminals and projections of neurons. See [Relations for neurons](#Relations_for_neurons) for details. -e.g. -TBA +### Taxon constraints -The following should be recorded using an annotation_property axiom: +We can record taxon specificity of terms using -never_in_taxon +**'in taxon'** *some* -e.g. +For example, the term 'alpha motor neuron' refers to a type of motor neuron that innervates skeletal muscle in vertebrates, so we can record -TBA +'alpha motor neuron' subClassOf **'in taxon'** *some* 'Vertebrata ' +Other relation are available for recording taxon constraints (details TBA) Further reading: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC3098089/ -Inference of taxon constraints from other ontologies: - -Relationships to taxon constrained terms from other ontologies can result in inferred taxon constraints. - -Examples: - - Inference of Taxon constraints from taxon constrainted makers: - - Inference of Taxon constraints from taxon constrainted anatomy: ### Recording function We record cellular function by linking to GO biological process terms using the relation (objectProperty) **'capable of'** -e.g. 'capable of' some 'androgen secretion' +e.g. 'hilus cell of ovary' **'capable of'** *some* 'androgen secretion' ### Recording developmental lineage We record developmental lineage relationships between cell types using **develops from**, or where we are sure there are not intermediates between the related cells, by using **'develops directly from'** +For example: + +'leukocyte' **develops from** *some* 'hematopoietuc stem cell' + ### Relations for neurons #### Synaptic connectivity From d72399c71b3176d907f43c9fdaffbe284146cab8 Mon Sep 17 00:00:00 2001 From: David Osumi-Sutherland Date: Thu, 1 Oct 2020 15:05:56 +0100 Subject: [PATCH 5/5] Update relations_guide.md --- documentation/relations_guide.md | 42 +++++++++++++++----------------- 1 file changed, 19 insertions(+), 23 deletions(-) diff --git a/documentation/relations_guide.md b/documentation/relations_guide.md index f91f5271a..1da2cc3c5 100644 --- a/documentation/relations_guide.md +++ b/documentation/relations_guide.md @@ -20,7 +20,7 @@ Some cells, most obviously neurons, only have some parts in the anatomical struc We also have a dedicated set of relations for recording the location of synaptic terminals and projections of neurons. See [Relations for neurons](#Relations_for_neurons) for details. -### Taxon constraints +## Taxon constraints We can record taxon specificity of terms using @@ -34,13 +34,13 @@ Other relation are available for recording taxon constraints (details TBA) Further reading: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC3098089/ -### Recording function +## Recording function We record cellular function by linking to GO biological process terms using the relation (objectProperty) **'capable of'** e.g. 'hilus cell of ovary' **'capable of'** *some* 'androgen secretion' -### Recording developmental lineage +## Recording developmental lineage We record developmental lineage relationships between cell types using **develops from**, or where we are sure there are not intermediates between the related cells, by using **'develops directly from'** @@ -48,9 +48,9 @@ For example: 'leukocyte' **develops from** *some* 'hematopoietuc stem cell' -### Relations for neurons +## Relations for neurons -#### Synaptic connectivity +### Synaptic connectivity To record neuron to neuron or motor neuron -> target cell connectivity use. Use these relations sparingly where connectivity is key to definition, e.g. motor neuron types defined by the type of muscle cell they synapse to. @@ -62,33 +62,29 @@ To record connection between a neuron and a region it innervates we have a numbe ![image](https://user-images.githubusercontent.com/112839/94337631-e0a83300-ffe3-11ea-8f13-ac8a484a5fb3.png) -Historically we have used *has synaptic terminal in* +Historically we have used **has(pre/post)synaptic terminal in** to record this. However, these relations are defined as being true when a single synapse is present in a region. As these relations are also used with data, this turns out to be too sensitive to biological and experimental noise. We therefore now prefer the more specific relations for defining classes - defined in terms of functionally significant synaptic inputs/output in a region. -### Recording cell markers +## Recording cell markers -has part - . has plasma membrane part - -expresses - . has plasma membrane part - - Absence of a marker - lacks_plasma_membrane_part - -#### A note on cell markers and taxon constraints +Only markers that are necessary to define a cell type should be recorded. + +NOTE: The details of how and when we record cell markers are are in flux. If in doubt, ask an editor. + +The most commonly used relation for recording markers is **'has plasma membrane part'**. This is used for recording cell surface markers, especially in immune cells. We also have subproperties **has low plasma membrane amount** and **has low plasma membrane amount'**, used to the same end. In each case, a term from the protein ontology or a protein complex term from GO is used as the object of the relation: + +e.g. alpha-beta T cell EquiavlentTo 'T cell' *and* **'has plasma membrane part'** *some* 'alpha-beta T cell receptor complex' -If you choose a cell specific marker +Absence of a marker can be recorded using **lacks_plasma_membrane_part**. -#### A note on when to record cell markers -### Recording cell shape or other morphological qualities +## Recording cell shape or other morphological qualities -e.g. erythrocyte bearer_of some biconcave +e.g. erythrocyte subClassOf **bearer of** *some* biconcave -### Recording cellular qualities (eg. ploidy, nuclear number) +## Recording cellular qualities (eg. ploidy, nuclear number) -e.g. 'enucleate erythrocyte' EquivalentTo erythrocyte and ('bearer of' some anucleate) +e.g. 'enucleate erythrocyte' EquivalentTo erythrocyte *and* **'bearer of'** *some* anucleate