Finding and Querying Networks

Last updated: June 30th, 2017

The public NDEx site includes a number of networks that are marked as "PUBLIC" and are therefore accessible without signing in to a user account. Public networks can be found, viewed, and queried but you cannot create new networks, upload, or download without signing in.

IMPORTANT NOTE: The screenshots in all our user manuals are for training purposes and might be different from the actual screen output you will obtain when working on the current NDEx version deployed to our public server.

Some Public Networks on NDEx

  • The collection of Nature-curated NCI Pathway Interaction Database networks
  • OpenBEL (BEL format)
    • BEL small corpus
    • BEL large corpus
  • A collection of lipid metabolic pathways from LIPID MAPS (BioPAX3 format)

Searching for Networks Based on Title and Description

  • Search text can be entered directly into the navigation bar.
  • The current search mode is displayed to the left of the search box and defaults to "Networks".
  • Try running an empty search: just click magnifying glass or press enter to search.
  • The system will return all the networks you have access to! The network search results page displays all the hits and provides basic info about the networks, such as format, owner, last modification date etc.
  • If you hover on a network name, a pop up window will display the network description.
  • Only public networks are returned in a search; a private network will only appear in a search if you have been granted access to that network by its administrator or if you are the owner of that network.

View a Network Found in a Search

  • To inspect a network, click its name.
  • The network page will open up: information about the network is displayed on the left (red line) and includes the counts of nodes and edges for the entire network, date of creation, UUID and links to the network administrators profiles. Network description and literature reference are also displayed if available. Toward the bottom, some network properties are also displayed
  • If a network is larger than 300 edges, no graphical display will be rendered and the main part of the page will only display a table containing a sample of 500 randomly selected edges.
  • The table has 3 tabs: one for Edges, one for Nodes and the last one for the Provenance history (red arrow).
  • On the right hand side, there is a link for the Advanced query interface (green arrow).

Advanced Search Capabilities

Starting from NDEx v1.3, a stand-alone Apache SOLR service is used to implement Lucene Indexing. Advanced search capabilities allow users to search (and query) networks using keywords or other network attributes. A complete list of the indexed attributes is available in the tables below.

SEARCH: Indexed network attributes

Note that some attributes below are highlighted in blue text. These are NDEx internal attributes and are not encoded in the CX networkAttributes aspect. In the case of attributes like ‘edgeCount', they are calculated values.

Attribute Description Type
labels Users can use this property to specify custom tags freetext
methods Description of methods used to generate the network freetext
networkType Specifies the type of interactions in the network (protein, genetic, chemical) freetext
organism Description of organisms associated with the network freetext
disease Description of diseases associated with the network freetext
tissue Description of organs, tissue types, cell types, and cell lines associated with the network freetext
rightsHolder Describes the holder of the rights to the network – same as dc:rightsHolder freetext
rights Description of the rights asserted by the rightsHolder – same as dc:rights. The indexing of this general field should aggregate not only the actual ‘rights’ attribute but the more specific rights attributes, such as copyright or license. In that way, network authors do not need to state information redundantly freetext
author Individuals responsible for creating the content of the network – see dc:contributors freetext
name Network title freetext
description A description of the network, its content and data sources freetext
version Network version string
creationTime Time at which the network was created date
modificationTime Time at which the network was last modified date
edgeCount Number of edges in the network integer
nodeCount Number of nodes in the network integer
visibility NDEx network visibility: Public, Private string
uuid This corresponds to the externalId ndexStatus element attribute string
owner This corresponds to the user account name that owns a network string

SEARCH: Indexed node attributes
Attribute Description Type
nodeName Name of node freetext
represents string
alias Alternative identifiers for node string

QUERY: Indexed node attributes
Attribute Description Type
nodeName Name of node freetext
represents string
alias Alternative identifiers for node string

Advanced Search Examples

  • You can search for networks using keywords: type "metabolism" in the search bar and hit enter.
  • The system returns all networks that have the term "metabolism" in their name, description or associated to its nodes and/or edges.

  • You can also use wildcards: if you type "met*" the system will return all networks that have metabolism, metabolic, methionine, metformin and the met gene referenced in their title, description, nodes or edges. In the screenshot below, the last network has the keyword "metabolic" in its description (red circle).

  • However, if you tipe "name:met*", the system will only return networks that have those terms in their name: in the screenshot below, you can see that the network called Biosynthesis of Sterols is missing as the term "metabolic" is contained in its description.

  • If you know the unique identifier (UUID) of a network, you can search NDEx using this information.
  • type "uuid": in the search bar followed by the UUID of the network you are looking for.

  • You can search networks specifying a range of values: type "nodeCount:[11 TO 79]" in the search bar and press enter.
  • The system will return all the available networks that have between 11 and 79 nodes.

  • You can search networks based on when they were created or last modified.
  • type "creationTime:[2016-02-26T00:00:01Z TO 2016-02-27T23:59:59Z]" and press enter.
  • The system will return all available networks that were created between 12:01 am on Feb 26, 2016 and 11:59 pm on Feb 27, 2016. Do not confuse creation time with last modification time (red circle below).

  • You can search networks owned by a certain account. To do this, typea range of values: type owner: followed by the user account name. For example, try to search for owner:ndextutorials
  • The system will return all the available networks that are owned by the account "ndextutorials".
  • You can perform complexes searches combining several keywords and properties. For example, let's search all networks that:
    • have the keyword ‘metabolism" in their name
    • have no more than 5000 edges
  • Type "name:metabolism AND edgeCount:[1 TO 5000]" and press enter to get the results

  • Starting from NDEx 2.1, you can decide to perform Search Term Expansion on genes and proteins. This feature expands your search term(s) to all its possible synonims and runs a search using all of them. This makes sure that you will find all networks that have your genes/proteins of interest even if the genes/proteins of interest are listed using an uncommon synonim or an alternative identifier.
  • To use the Search Term Expansion feature, just type your genes/proteins of interest in the search box and mark the "Search Term Expansion" checkbox below:

  • Please not that Search Term Expansion is not compatible with Lucene Syntax.
  • For more information and details about Lucene Syntax, please refer to the Lucene Documentation.

Query for a Neighborhood in a Network

Although when viewing a network the table only displays a sample of 500 Edges, any queries you perform will be always executed on the entire network. NDEx allows users to run 2 types of queries: Simple or Advanced.

Simple Query

In the text box in the query controls, you can enter terms to query in the network and the system will find a neighborhood around all nodes that reference those terms. The resulting network is defined by the "depth" of the query: 1-Step finds only the immediate neighbors of the nodes. For example:

  • Type "cox2", select a depth of "1-step" and click the "Run Query" button.

  • The retrieved network is now a small neighborhood specified by the "cox2" term, consisting of "16 nodes" and "16 edges" (red arrow). All the retrieved nodes and edges can be inspected in the table and additional filtering is possible using the text box at the top of each table's column (green arrow).
  • Click "Back to Original Network" to clear your query result and go back to the previous screen

Advanced Query

Starting from NDEx v1.2, the Advanced Query feature allows users to query a network based on properties associated to its nodes and/or edges.

  • Click "Back to Original Network" and then click the Advanced Query link on the right hand side.

As pictured below, the Advanced Query interface will show up, allowing you to search the network by filtering nodes and edges based on their properties.

  • Available properties can be selected using the drop down menus. Properties can also be identified by scrolling to the right in the Edges/Nodes table. You can select as many properties as you want on both Edges and Nodes. Properties The property values you type in text box to the right are case-insensitive but must be a complete match
  • For example, choose "Disease" in the property drop down menu and type "Breast Neoplasms" in the value text box, then click "Run Query".

  • The system will return a network consisting of nodes and edges where the Edges have the Disease property defined as "Breast Neoplasms". The returned subnetwork will not include any nodes and edges whose property values have a partial match… so, in the example above, no nodes and edges will be included that have a disease property defined as "Neoplasm" or "Breast Cancer" or " Breast Tumor". Also, if you type "Breast Neoplasm" instead of "Breast Neoplasms", no results will be returned as the property value is you typed is not an exact match.

  • In addition, the Advanced Query also allows you to query a network by predicate: click the "Back to Original Network" button, select "ndex:predicate" in the property drop down menu and type "positive_correlation" in the value text box and then click "Run Query":

This time the system will return a neighborhood consisting of 74 edges whose predicate value equals to "positive_correlation".