RBGL shortest path always giving path length NA and path_details NA

[mharoun20]

I tried to implement the RBGL code from the paper:

library(RBGL) # install from Bioconductor

Convert a sample Cytoscape network to a graph object

openSession() g <- createGraphFromNetwork()

Identify start and finish nodes (styling is optional)

start <- "YNL216W" finish <- "YER040W" setNodeBorderWidthBypass(c(start,finish), 20) setNodeBorderColorBypass(start,"#00CC33") setNodeBorderColorBypass(finish,"#CC00CC")

Use RBGL to perform shortest path calculation

shortest <- sp.between(g, start, finish) shortest$'YNL216W:YER040W'$length

But I always get length and path details of NAs and path length of 1. I tried it on multiple networks include the sample networks at cystoscope. I am using RStudio Version 1.4.1106 with R version 4.2.2 on Windows 10. Any idea why this is happening?

[AlexanderPico]

When I tried running this notebook, I ran into a duplicate edge error in the step:

g <- createGraphFromNetwork()
Error in igraph::igraph.to.graphNEL(ig) : 
  multiple edges are not supported in graphNEL graphs

Somehow, a duplicate edge was inserted into the sample network since the vignette was written. I've inserted a couple lines to simply delete it just after openSession():

selectEdges("YPL248C (pp) YML051W","name")
deleteSelectedEdges()

After that, the rest of the notebook ran fine and produced the expected result. Does this address your issue?

[szhorvat]

Somehow, a duplicate edge was inserted into the sample network since the vignette was written.

Perhaps it was always there, but it wasn't detected. igraph only added a check for multi-edges in version 1.4.0. Conversions of multigraphs to graphNEL before that version should be considered invalid. See https://github.com/igraph/rigraph/issues/575

Generally, graphNEL appears to support only simple graphs, but this issue is not discussed in detail in its documentation. There is a passing reference mentioning that multi-edges won't be handled properly, and there is no statement on self-loops at all. See also https://github.com/Bioconductor/graph/issues/17 where I asked about this and waiting to receive a response.