Merge pull request Reed-CompBio#103 from livj4711/domino

Add DOMINO pathway reconstruction algorithm

.github/workflows/test-spras.yml

@@ -81,6 +81,7 @@ jobs:
         docker pull reedcompbio/pathlinker:latest
         docker pull reedcompbio/meo:latest
         docker pull reedcompbio/mincostflow:latest
+        docker pull reedcompbio/domino:latest
     - name: Build Omics Integrator 1 Docker image
       uses: docker/build-push-action@v1
       with:
@@ -126,6 +127,15 @@ jobs:
         tags: latest
         cache_froms: reedcompbio/mincostflow:latest
         push: false
+    - name: Build DOMINO Docker image
+      uses: docker/build-push-action@v1
+      with:
+        path: docker-wrappers/DOMINO/.
+        dockerfile: docker-wrappers/DOMINO/Dockerfile
+        repository: reedcompbio/domino
+        tags: latest
+        cache_froms: reedcompbio/domino:latest
+        push: false
 
   # Run pre-commit checks on source files
   pre-commit:

config/config.yaml

@@ -54,6 +54,7 @@
                 run2:
                     b: [2]
                     g: [3]
+
         - name: "meo"
           params:
                 include: true
@@ -71,6 +72,13 @@
                     flow: [1] # The flow must be an int
                     capacity: [1]
 
+        - name: "domino"
+          params:
+                include: true
+                directed: false
+                run1:
+                    slice_threshold: [0.3]
+                    module_threshold: [0.05]
 
  # Here we specify which pathways to run and other file location information.
  # DataLoader.py can currently only load a single dataset
@@ -88,7 +96,7 @@
      -
        label: data1
        # Reuse some of the same sources file as 'data0' but different network and targets
-       node_files: ["sources.txt", "alternative-targets.txt"]
+       node_files: ["node-prizes.txt", "sources.txt", "alternative-targets.txt"]
        edge_files: ["alternative-network.txt"]
        other_files: []
        # Relative path from the spras directory

config/egfr.yaml

@@ -62,6 +62,16 @@ algorithms:
           - 3
         rand_restarts:
           - 10
+  -
+    name: domino
+    params:
+      directed: false
+      include: true
+      run1:
+        slice_threshold:
+          - 0.3
+        module_threshold:
+          - 0.05
 datasets:
   -
     data_dir: input

docker-wrappers/DOMINO/Dockerfile

@@ -0,0 +1,11 @@
+# DOMINO wrapper
+# https://github.com/Shamir-Lab/DOMINO
+FROM python:3.7
+
+RUN pip install domino-python==0.1.1
+
+# DOMINO requires data files in hard-coded locations
+RUN cd /usr/local/lib/python3.7/site-packages/src/data && \
+    wget https://raw.githubusercontent.com/Shamir-Lab/DOMINO/master/src/data/ensg2gene_symbol.txt && \
+    wget https://raw.githubusercontent.com/Shamir-Lab/DOMINO/master/src/data/ensmusg2gene_symbol.txt && \
+    wget https://raw.githubusercontent.com/Shamir-Lab/DOMINO/master/src/data/graph.html.format

docker-wrappers/DOMINO/README.md

@@ -0,0 +1,22 @@
+# DOMINO Docker image
+
+A Docker image for [DOMINO](https://github.com/Shamir-Lab/DOMINO) that is available on [DockerHub](https://hub.docker.com/repository/docker/reedcompbio/domino).
+
+DOMINO outputs multiple active modules, which SPRAS combines into a single pathway.
+It is [non-deterministic](https://github.com/Shamir-Lab/DOMINO/issues/5) and cannot be made deterministic with a seed.
+
+To create the Docker image run:
+```
+docker build -t reedcompbio/domino -f Dockerfile .
+```
+from this directory.
+
+To inspect the installed Python packages:
+```
+winpty docker run reedcompbio/domino pip list
+```
+The `winpty` prefix is only needed on Windows.
+
+## TODO
+- Resolve upstream ValueError with small inputs https://github.com/Shamir-Lab/DOMINO/issues/11
+- Use cache or reuse slices files from previous runs on the same network

input/README.md

@@ -11,11 +11,11 @@ All other columns specify additional node attributes such as prizes.
 Any nodes that are listed in a node file but are not present in one or more edges in the edge file will be removed.
 For example:
 ```
-NODEID	prize	sources	targets
-A	1.0		True
-B	3.3	True	
-C	2.5		True
-D	1.9	True	True
+NODEID	prize	sources	targets	active
+A	1.0		True	True
+B	3.3	True		True
+C	2.5		True	True
+D	1.9	True	True	True
 ```
 
 A secondary format provides only a list of node identifiers and uses the filename as the node attribute, as in the example `sources.txt`.
@@ -49,6 +49,7 @@ The files are originally from the [Temporal Pathway Synthesizer (TPS)](https://g
 They have been lightly modified for SPRAS by lowering one edge weight that was greater than 1, removing a PSEUDONODE prize, adding a prize of 10.0 to EGF_HUMAN, and converting all edges to undirected edges.
 The only source is EGF_HUMAN.
 All proteins with phosphorylation-based prizes are also labeled as targets.
+All nodes are considered active.
 
 If you use any of the input files `tps-egfr-prizes.txt` or `phosphosite-irefindex13.0-uniprot.txt`, reference the publication
 

input/node-prizes.txt

@@ -1,3 +1,3 @@
-NODEID	prize
-A	2
-C	5.7
+NODEID	prize	active
+A	2	true
+C	5.7	true