diff --git a/docker-wrappers/MinCostFlow/README.md b/docker-wrappers/MinCostFlow/README.md
index 014a3d75..5d585eea 100644
--- a/docker-wrappers/MinCostFlow/README.md
+++ b/docker-wrappers/MinCostFlow/README.md
@@ -36,3 +36,7 @@ docker run -w /data --mount type=bind,source=/${PWD},target=/data reedcompbio/mi
 This will run MinCostFlow on the test input files and write the output files to the root of the `spras` repository.
 
 Windows users may need to escape the absolute paths so that `/data` becomes `//data`, etc.
+
+## Versions
+- v1: Initial version. Handles undirected edges only.
+- v2: Updated to handle both directed and undirected edges.
\ No newline at end of file
diff --git a/spras/mincostflow.py b/spras/mincostflow.py
index b0a47cc7..3b25fdc6 100644
--- a/spras/mincostflow.py
+++ b/spras/mincostflow.py
@@ -55,7 +55,7 @@ def generate_inputs(data, filename_map):
         edges = convert_undirected_to_directed(edges)
 
         # creates the edges files that contains the head and tail nodes and the weights after them
-        edges.to_csv(filename_map['edges'], sep='\t', index=False, columns=["Interactor1", "Interactor2", "Weight"],
+        edges.to_csv(filename_map['edges'], sep='\t', index=False, columns=["Interactor1", "Interactor2", "Weight", "Direction"],
                      header=False)
 
     @staticmethod
@@ -112,7 +112,7 @@ def run(sources=None, targets=None, edges=None, output_file=None, flow=None, cap
             command.extend(['--capacity', str(capacity)])
 
         # choosing to run in docker or singularity container
-        container_suffix = "mincostflow"
+        container_suffix = "mincostflow:v2"
 
         # constructs a docker run call
         out = run_container(container_framework,
@@ -140,9 +140,7 @@ def parse_output(raw_pathway_file, standardized_pathway_file):
         """
         Convert a predicted pathway into the universal format
 
-        Although the algorithm constructs a directed network, the resulting network is treated as undirected.
-        This is because the flow within the network doesn't imply causal relationships between nodes.
-        The primary goal of the algorithm is node identification, not the identification of directional edges.
+        The algorithm constructs a mixed directional network, the resulting network is treated as mixed directional.
 
         @param raw_pathway_file: pathway file produced by an algorithm's run function
         @param standardized_pathway_file: the same pathway written in the universal format
@@ -151,8 +149,5 @@ def parse_output(raw_pathway_file, standardized_pathway_file):
         df = raw_pathway_df(raw_pathway_file, sep='\t', header=None)
         if not df.empty:
             df = add_rank_column(df)
-            # TODO update MinCostFlow version to support mixed graphs
-            # Currently directed edges in the input will be converted to undirected edges in the output
-            df = reinsert_direction_col_undirected(df)
             df.columns = ['Node1', 'Node2', 'Rank', "Direction"]
         df.to_csv(standardized_pathway_file, header=True, index=False, sep='\t')