updated documentation and small usability changes

Author: jjc2718

README.md

@@ -2,22 +2,52 @@
 
 Requires modules `networkx`, `numpy`, `scikit-learn`, and `argparse`.
 
-For a description of the Random Walk with Restart (RWR) algorithm, see the paper by Kohler et al. at [http://www.sciencedirect.com/science/article/pii/S0002929708001729](http://www.sciencedirect.com/science/article/pii/S0002929708001729).
+For a description of the Random Walk with Restart (RWR) algorithm, which
+this module implements, see the paper by Kohler et al. at
+[http://www.sciencedirect.com/science/article/pii/S0002929708001729](http://www.sciencedirect.com/science/article/pii/S0002929708001729).
 
-This module was initially created to run node removal experiments with two separate graphs, but the code in matrix\_main.py can easily be used to run the standard RWR algorithm on a single graph of any sort.
+## Overview
+
+This module can be used to run two types of experiments:
+
+- A standard random walk with restart from a set of seed nodes, as in the
+  Kohler et al. paper referenced above.
+- A random walk with restart, from a set of seed nodes, on a "tissue-specific"
+  network. The network is defined by a "low list" of nodes (i.e. genes) that
+  are not expressed in the tissue of interest. This is described in more
+  detail in our paper, which is currently in review.
+
+Examples of both experiments are described in more detail below.
 
 ## Running a random walk
 
-The matrix\_main.py script can be used to run a random walk. The syntax looks like:
+The run\_walker.py script can be used to run a random walk. The syntax looks like:
 
-`python matrix_main.py <input_graph> <seed> [-l <low_list>] [-r <remove_nodes>]`
+`python run_walker.py <input_graph> <seed> [-l <low_list>] [-r <remove_nodes>]`
 
 where the input graph is in edge list format, the seed is a list of nodes to
 start the random walk at, the optional low list is a list of nodes to down-weight
-for node removal experiments (as in the tissue-specific networks paper), and the
-optional node removal list is a list of nodes to remove completely from the graph.
+for node removal experiments, and the optional node removal list is a list of nodes
+to remove completely from the network.
+
+The script will write a tab-separated list of nodes and probabilities to stdout,
+where the probability number represents the probability that a random walk
+starting at the seed nodes will terminate at the given node.
+
+For more detail about the expected arguments, run `python run_walker.py -h`.
+
+## Examples
+
+To help you get up and running, a few simple examples are included in the `testdata`
+folder. To run a standard random walk experiment on a simple example network, run
+this command:
+
+`python run_walker.py testdata/test_network.ppi testdata/test_seed.txt`
+
+Or, to run a "tissue-specific" random walk experiment using the same
+simple example network, try:
 
-More thorough documentation to come.
+`python run_walker.py testdata/test_network.ppi testdata/test_seed.txt -l testdata/test_low_list.txt`
 
 ## Using the module
 

runSH.py

@@ -73,7 +73,7 @@ def main(argv):
 
     # run the experiments, and write a rank list to stdout
     wk = Walker(opts.input_graph, opts.low_list, remove_list)
-    wk.run_exp(seed_list, opts.restart_prob, 
+    wk.run_exp(seed_list, opts.restart_prob,
                opts.original_graph_prob, node_list)
 
 

run_walker.py

@@ -0,0 +1,5 @@
+0	1.5
+1	2.0
+2	NA
+3	1.0
+4	1.8

testdata/test_low_list.txt

@@ -98,7 +98,7 @@ def _generate_rank_list(self, p_t):
 
     def _calculate_next_p(self, p_t, p_0):
         """ Calculate the next probability vector. """
-        if self.tsg_matrix:
+        if self.tsg_matrix is not None:
             no_epsilon = np.squeeze(np.asarray(np.dot(self.tsg_matrix, p_t) *
                                     (1 - self.og_prob)))
             epsilon = np.squeeze(np.asarray(np.dot(self.og_matrix, p_t) *
@@ -151,7 +151,7 @@ def _build_matrices(self, original_ppi, low_list, remove_nodes):
                                                   og_not_normalized, low_list)
             self.tsg_matrix = self._normalize_cols(tsg_not_normalized)
         else:
-            self.tsg_matrix = []
+            self.tsg_matrix = None
 
 
     def _tsg_matrix(self, original_graph, og_matrix, low_list):