WISP / Discussion / General Discussion: Error running simple

I am attempting to run the simple_example for the lastest version of wisp, but I am getting the following error:

# Calculating paths...
#       Calculating the shortest path between any of the specified sources and any of the specified sinks...
Traceback (most recent call last):
  File "../wisp.py", line 1464, in <module>
    paths = GetPaths(correlation_matrix, sources, sinks, parameters, correlation_matrix_object.average_pdb.residue_identifiers_in_order)
  File "../wisp.py", line 1006, in __init__
    shortest_length, shortest_path = self.get_shortest_path_length(correlation_matrix, sources, sinks, G)
  File "../wisp.py", line 1109, in get_shortest_path_length
    short_path = networkx.dijkstra_path(G, source, sink, weight='weight')
  File "/cm/shared/apps/miniconda/lib/python2.7/site-packages/networkx/algorithms/shortest_paths/weighted.py", line 163, in dijkstra_path
    weight=weight)
  File "/cm/shared/apps/miniconda/lib/python2.7/site-packages/networkx/algorithms/shortest_paths/weighted.py", line 472, in single_source_dijkstra
    weight=weight)
  File "/cm/shared/apps/miniconda/lib/python2.7/site-packages/networkx/algorithms/shortest_paths/weighted.py", line 732, in multi_source_dijkstra
    cutoff=cutoff, target=target)
  File "/cm/shared/apps/miniconda/lib/python2.7/site-packages/networkx/algorithms/shortest_paths/weighted.py", line 817, in _dijkstra_multisource
    raise nx.NodeNotFound("Source {} not in G".format(source))
 networkx.exception.NodeNotFound: Source 10 not in G

I am using the version of wisp downloaded from http://sourceforge.net/projects/wisp-pathways/files/1.0/wisp-1.0.zip/download

extracted into ~/bin/wisp

I am running on CentOs version 6.9
with python 2.7.15
numpy 1.15.0
scipy 1.2.0
networkx 2.2
cPickle 1.71
multiprocessing 0.70a1

I have tried fiddling around a bit and setting the source and sink nodes to other values and I get the same 'networkx.exception.NodeNotFound: Source ### not in G' for any choice of source node

Last edit: Wesley 2019-04-01

Wesley - 2019-04-02

I think I may have tracked part of the error.
It looks like wisp.py is using

G=networkx.Graph(data=correlation_matrix

However, under networkx 2.2, the signature for networkx.Graph
is:

'networkx.Graph(self, incoming_graph_data=None, **attr)'

rather than 'networkx.Graph(self,data=None, **attr)'
as it is in some versions...

I am checking to see if updating this line will fix the bug

If you would like to refer to this comment somewhere else in this project, copy and paste the following link:

The bug fix seems to work and the shortest path is computed, however, a new error is coming up after 'Starting serial portion of path-finding algorithm':

# Calculating paths...
#       Calculating the shortest path between any of the specified sources and any of the specified sinks...
#           The shortest path has length 1.136358953726239
#      Identifying the cutoff required to produce 15 paths...
#          Testing the cutoff 1.136358953726239...
#                Starting serial portion of path-finding algorithm (will run for 5.0 seconds)...
Traceback (most recent call last):
  File "../wisp.py", line 1464, in <module>
    paths = GetPaths(correlation_matrix, sources, sinks, parameters, correlation_matrix_object.average_pdb.residue_identifiers_in_order)
  File "../wisp.py", line 1025, in __init__
    paths = self.remove_redundant_paths(self.get_paths_between_multiple_endpoints(cutoff_in_array, correlation_matrix, sources, sinks, G, parameters))
  File "../wisp.py", line 1149, in get_paths_between_multiple_endpoints
    paths.extend(self.get_paths_fixed_endpoints(cutoff, correlation_matrix, source, sink, G, parameters))
  File "../wisp.py", line 1235, in get_paths_fixed_endpoints
    find_paths_object.expand_growing_paths_one_step(paths_growing_out_from_source, full_paths_from_start_to_sink, cutoff, sink, G)
  File "../wisp.py", line 761, in expand_growing_paths_one_step
    node_neighbors=G.neighbors(paths_growing_out_from_source[i][-1])
  File "/cm/shared/apps/miniconda/lib/python2.7/site-packages/networkx/classes/graph.py", line 1251, in neighbors
    raise NetworkXError("The node %s is not in the graph." % (n,))
networkx.exception.NetworkXError: The node 10 is not in the graph.

Anyone have any ideas what is going on?

Wesley - 2019-04-02

It seems that the error is again do to a different signature.
Some searching reveals that this was written for networkx versions below 2.0. New versions of networkx (2.0 and up) have had several major changes, such as the signature of the 'Graph' constructor.

There are 2 uses of this constructor in 'wisp.py'
line 1216 and line 1001
in both cases, 'Graph(data=' needs to be replaced with 'Graph(incoming_graph_data=' when using networkx >= 2.0

Fixing these issues leads to a new bug, however:
On line 762, 'for j in range(len(node_neighbors)):' causes an error because new versions of networkx are using dictionary-keyiterator objects to track node_neighbors. This leads to the error:
TypeError: object of type 'dictionary-keyiterator' has no len()
The solution is to wrap 'node_neighbors' in 'list()' so

for j in range(len(node_neighbors)):
is replaced by
for j in range(len(list(node_neighbors))):

If you would like to refer to this comment somewhere else in this project, copy and paste the following link:

Wesley - 2019-04-02

The above fix seems to just open up a related problem...
it seems that previous versions of networkx use 'list' objects as return values for the 'neighbors' function, but newer versions use 'dictionary-keyiterator' objects.
So the solution is to wrap the result being fed into 'node_neighbors' in 'list()'
e.g.
line 761
node_neighbors=G.neighbors(paths_growing_out_from_source[i][-1])
becomes
node_neighbors=list(G.neighbors(paths_growing_out_from_source[i][-1]))

If you would like to refer to this comment somewhere else in this project, copy and paste the following link:

Wesley - 2019-04-02

After the above fix, another quirk of newer networkx versions was found.
Line 766:
temp[0]=temp[0]+G.edge[temp[-2]][temp[-1]]['weight']
needs to be changed to
temp[0]=temp[0]+G.edges[temp[-2],temp[-1]]['weight']
in order to accomadate the new syntax for edge (or edges in v>2.0) access

If you would like to refer to this comment somewhere else in this project, copy and paste the following link:

Wesley - 2019-04-02

SUCCESS!
After the above bug fixes, wisp.py is now sufficiently compatible with networkx >= 2.0 such that it can successfully run the 'simple-example' that comes with the installation.

Can anyone point me to who to contact to construct a patch / update / bug fix/ etc?

If you would like to refer to this comment somewhere else in this project, copy and paste the following link:

jdurrant - 2019-07-23

Wesley, please forgive me for my delay in getting back to you! My sourceforge emails were going to my old UCSD email address, but I'm now at the University of Pittsburgh. Wonderful that you were able to patch the code to work with the latest version of networkx. Would you mind emailing your updated code to me at durrantj at pitt dot edu? All the best.

If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
- Wesley - 2019-07-25
  
  Hi, I tried sending the code you, but I got a reply that the message was not delivered:
  
  Delivery has failed to these recipients or groups: Your message wasn't delivered due to a permission or security issue. It may have been rejected by a moderator, the address may only accept email from certain senders, or another restriction may be preventing delivery.
  
  i am going to attache the code here.
  
  I have just tested it on a new iMac workstation working with a fresh install of conda (4.6.14, via anaconda), python 2.7 (2.7.16), pip (19.0.3), and networkx (2.2)
  
  let me know if you need anything else.
  
  wisp.py
  
  If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
  - jdurrant - 2019-07-31
    
    Hi Wesley. Just wanted to let you know that I got your code. Much thanks for this contribution! I've gotten this version running on Python 3 as well. We're going to test it intnerally for a bit and then publish it for others to use. Can I include you in the CONTRIBUTORS.md file? All the best.
    
    If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
    - Wesley - 2019-08-01
      
      Yes, feel free to include me in the CONTRIBUTORS file if the patch checks
      out. That would be great! Hopefully it will help others. It's always
      frustrating when api changes in dependencies break previously functional
      packages and libraries.
      All the best,
      -Dr. Wesley Botello-Smith
      
      On Tue, Jul 30, 2019 at 5:06 PM jdurrant jdurrant@users.sourceforge.net
      wrote:
      
      Hi Wesley. Just wanted to let you know that I got your code. Much thanks
      for this contribution! I've gotten this version running on Python 3 as
      well. We're going to test it intnerally for a bit and then publish it for
      others to use. Can I include you in the CONTRIBUTORS.md file? All the best.
      
      Error running simple_example
      https://sourceforge.net/p/wisp-pathways/discussion/general/thread/55d7445d45/?limit=25#59d2/4e21/a9ef
      
      Sent from sourceforge.net because you indicated interest in
      https://sourceforge.net/p/wisp-pathways/discussion/general/
      
      To unsubscribe from further messages, please visit
      https://sourceforge.net/auth/subscriptions/
      
      alternate
      
      If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
      - jdurrant - 2019-09-13
        
        Hi Wesley. Wanted to let you know that the latest version is posted here: https://git.durrantlab.pitt.edu/jdurrant/wisp
        
        All the best.
        
        If you would like to refer to this comment somewhere else in this project, copy and paste the following link:

Douglas Houston - 2019-08-20

That's great! I'd be very interested in trying out this new version ...

If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
- jdurrant - 2019-09-13
  
  Hi Douglas. Wanted to let you know that the latest version is posted here: https://git.durrantlab.pitt.edu/jdurrant/wisp
  
  All the best.
  
  If you would like to refer to this comment somewhere else in this project, copy and paste the following link:

Error running simple_example

Weighted Implementation of Suboptimal Paths (WISP)

Forums

Help

Error running simple_example

Error running simple_example

Weighted Implementation of Suboptimal Paths (WISP)

Forums

Help

Error running simple_example document.SUBSCRIPTION_OPTIONS = { "thing": "topic", "subscribed": false, "url": "subscribe", "icon": { "css": "fa fa-envelope-o" } };

Error running simple_example