From NeuroLex

Revision as of 02:24, 27 October 2011 by Bandrow (Talk | contribs)
Jump to: navigation, search

Below are some instructions and tutorials about the structure of NeuroLex and how to use its features. But don't worry; we'd rather you contribute than get everything right. We'll be monitoring the pages and providing some curation to clean things up now and then.


Quick start guide

We recommend you view the brief video below, but to just get some basic wiki editing tips click here.

Click here for updated video:

To quickly create a new category page

Decide on a name. Let's say you pick "Eric Kandel Motor Neuron". Type: Kandel Motor Neuron

into your browser window (where the URLs go). Hit enter. Click the "create" tab at the top of the page, or the link in the body of the page that says "edit this page". Type in this page. Click "save". Your changes are now part of the NeuroLex.

We have also created a set of forms on the home page that lets you enter a term to create a generic page, a neuron, a brain region or a resource. When you use these forms, the superclass is automatically assigned.

To assign a parent category

Click the "edit" tab at the top of the page. Insert:


in the page to make this page recognizable as a neuron in the NeuroLex. The square brackets help the NeuroLex to know that you mean to make this information machine processable. Click "save". At the bottom of the page, you'll see a link the the "neuron" category page. Click this and look up your new neuron in the dynamic list of neuron types.

Once you are done, please delete your test neuron by clicking the "delete" tab at the top of the page.

Editing pages with forms

Some of our category pages provide simple forms to enter information about the category and unique, related properties.

Creating Tables

Creating Tables with category and property information is easy. Follow the steps below.

  1. Use your browser to type a URL with the appropriate page name. For example you could type Table, which creates a page called "Diseases Table"
  2. Then, click the link that says "edit this page"
  3. To make a table of all the diseases in Neurolex, you would enter the text below.
{{#ask: [[Category:Disease]]
| ?synonym
| ?definition
| ?externallySourcedDefinition
| limit=1000

In order to personalize this text, you can change the category name or any of the properties you would like to list.

We recommend you look at other table pages from the main page and click the edit button for more examples of creating tables.

Creating Hierarchies

Creating Hierarchies with category and property information is easy. Follow the steps below.

  1. Use your browser to type a URL with the appropriate page name. For example you could type Hierarchy, which creates a page called "Diseases Table"
  2. Then, click the link that says "edit this page"
  3. To make a hierarchy of all the diseases in Neurolex, you would enter the text below.
<categorytree mode=categories depth=20 showcount=true>Disease</categorytree>

In order to personalize this text, you can change the category name to display a different hierarchy of terms

We recommend you look at other hierarchy pages from the main page and click the edit button for more examples of creating hierarchies.

Best practices

Is something missing from the NeuroLex? Did you find an error? Do you disagree with a definition? Then get to work on the NeuroLex. Here's how it works:

Annotation properties

Because NeuroLex is meant to help us design more effective and powerful information systems for neuroscience, we are looking for some very specific things for a NeuroLex entry.

Each unique concept in the NeuroLex is called a "Category Page". For each category, we are looking for the following:

  1. A clear human-readable definition that would allow someone to be able to apply that concept unambiguously to data and that would allow our knowledge engineers to build that concept into one of our ontologies. The best way to construct the definition is to go back to Aristotle: A is a B which has C. In other words, A is a type of B which can be distinguished from other types of B by a set of properties, e.g., A Purkinje cell is a type of neuron that is characterized by a large, pear shaped cell soma and an extensively branching dendritic tree that is densely covered with dendritic spines etc. In this case, "Neuron" is the parent class and Purkinje cell is a subclass. As discussed below, the NeuroLex uses a special Wiki that lets you define classes and subclasses.
  2. Synonyms, lexical variants, abbreviations. In the NeuroLex, the definition is key; if something has a different definition, then it is a different concept, even if it has the same name. If it has a different name, but the same definition, then we consider it a synonym and will add it as a synonym to the category page, e.g., Purkinje cell is the name of the category page; Purkinje neuron is listed as a synonym.
  3. Source of the concept, e.g., another vocabulary, a textbook, an article and the external citation.
  4. The species in which the concept holds, if applicable. For example, the Frontal Lobe is not found in rodents.
  5. Any diagrams or pictures that would help clarify the definition
  6. Any other comments

Although NeuroLex may look like Wikipedia, it is actually built using a special extension to the Wikipedia platform, called the Semantic Media Wiki. The Semantic Media Wiki lets us create formal relationships between category pages and apply a standard set of annotations. For example, if you go to the category page for Neuron, you will see a list of neurons. All of these have been defined as subclasses of Neuron in the NIFSTD ontology.

To provide a consistent format for each category in the NeuroLex, we have defined a standard set of annotation properties for each of desired fields:

  1. Definition
  2. Synonym or Abbreviation
  3. Ext Source
  4. Species
  5. Image
  6. Comment

For a list of defined properties in NeuroLex, click here

Conventions for creating category pages

Neurolex is a Wiki where the community can leave their knowledge behind. But it is curated which means we will be looking over the content and commenting or changing as appropriate. In general, we will try to restrain ourselves, but there are some conventions that we will enfore:

1) Categories should be singular and not plural, e.g., mossy fiber not mossy fibers. In some cases, the singular is rarely used, e.g, meninx is singular for meninges. In that case, we list the plural as a synonym for the class. But the preferred label is singular.

2) Category names must be unique: In actuality, the name of the category is a unique alpha numeric identifier, but each category is assigned a human readable label and that label should also be unique. When you type in a label, the wiki ensures that it is unique within the neurolex. However, we know that there are many "homographs" in biology, that is, terms that can mean two different things, e.g., nucleus as brain region and nucleus as part of cell. If there are two or more legitimate use of a term, you must supply a label to the category that is distinguishable, e.g., brain nucleus; cell nucleus. However, any occurrence of nucleus may be either, so it is important that humans or other agents know that there are multiple definitions of the term "nucleus". We therefor add "nucleus" as a synonym to both classes.

3) Category names should be complete in and of themselves. Many times when listing categories under a parent, we provide only a partial name, e.g., oblique may be a category under dendrite. Obviously, the oblique refers to an "oblique dendrite". If users are only looking at the hierarchy, this isn't a problem. However, if you are searching for a term and "oblique" comes up, there is no way from the label to know that it is really an "oblique dendrite". So the correct label for this concept should be "oblique dendrite".

4) Special case: Naming of neurons: The NIF cell working group along with the INCF Neuron Registry Task Force is trying to come up with conventions for naming neurons. Right now, we are recommending that each neuron be identified first by the brain region where it was identified followed by other properties that distinguish it, e.g., Cerebellum Purkinje neuron; Cerebellum granule cell. We know that some cell types are not confined to a particular brain region and will make recommendations about these cells in the future.

Conventions for assigning supercategories

1) All subcategories (aka children) should be related to the supercategory (aka parent class) by the "is a type of" relationship. That is, you should be able to translate the child-parent relationship into the following" Child class is a type of Parent class", e.g., "Neuron is a type of Cell". If this statement is not true, do not use the Super/sub category relationship. A common error is to place parts of a class as children of a class, e.g., putamen as a subclass of striatum. This would lead to the statement: Putamen is a type of Striatum" which is not true. Rather, use the "part of" relationship to link putamen to striatum, i.e., "Putamen is a part of Striatum".

2) It's not necessary to assign a superclass, but it makes bookkeeping easier and also provides some customization for different types of classes. For example, NIF generates a custom form for brain regions and neurons once a class is placed in these categories. Inititally, NIF created a lot of artifical classes to ensure that the "is a" relationship could be enforced, hence the large number of "Regional part of..." "Cytoarchitectural part of..." classes. We have found these clunky and difficult to maintain, and so are gradually removing them from Neurolex. For all brain regions, we recommend placing the class under the "Regional part of brain" or "Regional part of spinal cord" as appropriate. That way, the appropropriate form is generated.

3) Unclassified: For things we don't know about, e.g., impulsivity, we put them in a catch all class called "Unclassified". Again, this is largely for book keeping purposes. NIF curators go through this class and assign parents where appropriate. We also use a different set knowledge model in the underlying ontology when we don't know exactly what the thing is but we know what it is related to.