Create a lists of clustered nodes in a network

[TigerCole]

Hi…

I have a CSV file that lists the relationships between nodes in a whole network that contains sub-networks. An example of what may look like is the image below.

The network CSV looks like this:

Src	Dst	Label
A	B	has
A	E	has
B	C	has
B	H	has
H	I	has
C	G	has
C	D	has
L	S	has
O	M	has
J	K	has
K	N	has
N	J	has
P	R	has
R	P	has
P	Q	has
Q	R	has
C	H	has

What I need to do is extract clustered nodes into a list so that the final output table of the network looks something like this:

Network	Nodes
Net1	S,L
Net2	A,E,B,C,D,G,H,I
Net3	M,O
Net4	J,K,N
Net5	P,Q,R

I there a way to create the network in KNIME and then create this table of clustered nodes?

Thanks

tC/.

[aworker]

Hi @TigerCole

The best way of dealing with this question is to use the Network plugin available in KNIME:

Please find a possible solution here below:

20211006 Pikairos Create a lists of clustered nodes in a network.knwf (339.5 KB)

Hope this helps.

Best

Ael

PS: Complementary workflows and information about network graphs can be found in the KNIME hub:

KNIME Hub Graph Network examples

[TigerCole]

Hi @aworker …

This is exactly what I have been trying to do. Thank you for the example workflow. it has saved me days of work.

tC/.

[aworker]

HI @TigerCole

I’m glad it worked. Your kind feedback is my best reward and thanks for validating the solution !

Just a comment about this solution. As you will notice, the loop iterates as many times as nodes in the graph. This is fine with graphs with a reasonable number of nodes. However, if your network is made of millions of nodes but much less sub-graphs, an optimized alternative solution could be implemented with a recursive loop. I just wanted to let you know in case you have a huge network. Please get back in touch if this is the case for you. I’ll be happy to help further.

Best wishes,

Ael

[TigerCole]

Hi @aworker …

I did notice the looping but the flow finished in a few seconds… I am sure I will be posting on the forum again about networks. It is something that I am encountering more and more and it may make some of the things I need to do a lot more efficient, especially where data elements are closely related.

Thanks again for your help.

tC/.