Dear all, I have about 5k pairs of genes, predicted to have interactions from STRING website. The data structure I got from STRING is like this:
#node1 node2 node1_string_internal_id node2_string_internal_id node1_external_id node2_external_id neighborhood_on_chromosome gene_fusion phylogenetic_cooccurrence homology coexpression experimentally_determined_interaction database_annotated automated_textmining combined_score
UPF3B CASC3 4436211 4435465 9606.ENSP00000276201 9606.ENSP00000264645 0 0 0 0 0.062 0.993 0.900 0.808 0.999
SNRPF SNRPA 4435728 4433858 9606.ENSP00000266735 9606.ENSP00000243563 0 0 0 0 0.600 0.975 0.900 0.191 0.999
...
I want to use Cytoscape to visualize their network. After importing the data, the original network is like the following:
I only want all the nodes directly or indirectly connected with both gene A and gene B. So in this example, I only want gene A, B, C, E and F. Gene H, I and G are not wanted. Is there any existing tool or function in Cytoscape to realize it? Or do I need to write scripts to do the filtering? (I guess it will be very difficult and evolve lots of algorithms.
Thank you in advance.
Best, Yingzi
•
written
by
Yingzi Zhang
1 answer
Hi Yingzi, I have a couple of answers for you. First, you could just use the stringApp from within Cytoscape and provide it with genes A and B and you'll get exactly the network you want. However, if you want to use text file you downloaded from string-db, you could also load the entire network in Cytoscape, select gene A and gene B (either manually or using the filter tab) then select the first neighbors of those two nodes (Select->Nodes->First neighbors of selected nodes->Undirected). Now, you have all of the genes you are interested selected. Next, do a File->New Network->Selected Nodes, All Edges . That will create a network with only your selected nodes and the edges between them.
-- scooter
Log in to answer this question.