py2neo query subgraph from Neo4j

2
votes

I want to query a subgraph from a large graph data base with a given source. Say I want the top 25 neighbors of a given node, along with all the relations among this set of nodes, how should my query be?

At the moment I have:

MATCH (source {label:'source'}) -[:relation]-> (neighbors)
RETURN source,neighbors
LIMIT 25

This works in the neo4j browser, returning the 26 nodes as well as all the existing relations among those nodes. However, when I try execute the same query via py2neo:

py2neo.cypher.execute(query)

it only returns 26 nodes along with the 25 direct edge connections between the source and the 25 neighbors, which makes sense. But I wonder why there is a difference between the browser result versus the py2neo result. And how I can achieve the same result with all the edge connections returned.

In general, I would like to know the following smaller questions:

  • how to append a single node to a list of nodes in neo4j? e.g. nlist = neighbors + node

  • how to return all the relationships between two sets of nodes? e.g. return (a in nlist) -[:relation]-> (b in nlist)

Edits:

To visualize, I want the resulting graph to be something like this

subgraph

rather than a star graph like this

star

Thanks for any comments.

1
neo4jcypherpy2neo
Are you visualizing the results in Neo4j Browser using the graph visualization mode? If true, try changing to text mode for more precision representation. – Bruno Peres
Also, can you explain a bit more about the difference between results shown in neo4j browser and the result via py2neo? – Bruno Peres
Indeed I was viewing in the graph mode. And when I change to text mode it looks like I get the same results as I get from py2neo I believe. I have added a few more edits to illustrate what I mean by the difference. – thomas zhang
I believe the answer for your first question can be MATCH (source {label:'source'}) -[:relation]-> (neighbors) WITH collect(source) + collect(neighbors) as all UNWIND all AS nodes RETURN nodes. Right? – Bruno Peres
Great! This is exactly what I needed to start, with some extra tweaking and I find a good answer for the second question from here:stackoverflow.com/questions/25203804/… Would you like to reply via an answer and I can edit and accept this as a solution? Thanks for the input, this is a very good example for using collect I guess. – thomas zhang

1 Answers

1
votes

About the difference between Neo4j Browser and py2neo result: probably you are using the graph visualization mode with the option "Connect result nodes" enable. Try disabling (below image) it or change the visualization mode to "Text", for example.

Graph visualization options

About the Cypher query, you can use collect() to merge source and neighbors:

MATCH (source {label:'source'})-[:relation]->(neighbors)
WITH collect(source) + collect(neighbors) as all
UNWIND all AS nodes
RETURN nodes

In order to return the all the edges between two sets of nodes (in this case from set to itself), refer to this post:Neo4j, get all relationships between a set of nodes

Then the subgraph is returned with the Cypher query

MATCH (source {label:'source'})-[:relation]->(neighbors)
WITH collect(distinct source) + collect(neighbors) as all
UNWIND all AS nodes
MATCH (nodes)-[:relation]->(nei)
WHERE nei in nodes
RETURN nodes,nei