Staging (#42)

Co-authored-by: The Technocrat <[email protected]>

.gitignore

@@ -25,6 +25,7 @@ share/python-wheels/
 .installed.cfg
 *.egg
 MANIFEST
+data
 
 # PyInstaller
 #  Usually these files are written by a python script from a template

client.py

@@ -30,6 +30,52 @@ def generate_diffdock_instructions(
     }
     return json.dumps(instructions)
 
+#TODO: #26 introduce guidance on volume mounting, especially mounting multiple volumes
+def generate_vina_instructions(
+    debug_logs=True,
+    protein="1a30/1a30_protein.pdb",
+    ligand="1a30/1a30_ligand.sdf",
+    output="1a30/1a30_scored_vina.sdf.gz",
+    cnn_scoring="none",
+    modifier="score_only",
+) -> dict:
+    instructions = {
+        "container_id": "gnina/gnina:latest",
+        "debug_logs": debug_logs,
+        "short_args": {"v": "/home/ubuntu/casf-2016:/inputs"},
+        "long_args": {"gpus": 0},
+        "cmd": (
+            "gnina -r"
+            f" /inputs/{protein} -l /inputs/{ligand} -o"
+            f" /inputs/{output}"
+            f" --autobox_ligand /inputs/{protein} --cnn_scoring {cnn_scoring} --exhaustiveness 64"
+            f" --{modifier}"
+        ),
+    }
+    return json.dumps(instructions)
+
+def generate_gnina_instructions(
+    debug_logs=True,
+    protein="1a30/1a30_protein.pdb",
+    ligand="1a30/1a30_ligand.sdf",
+    output="1a30/1a30_scored_gnina.sdf.gz",
+    cnn_scoring="rescore",
+    modifier="score_only",
+) -> dict:
+    instructions = {
+        "container_id": "gnina/gnina:latest",
+        "debug_logs": debug_logs,
+        "short_args": {"v": "/home/ubuntu/casf-2016:/inputs"},
+        "long_args": {"gpus": 0},
+        "cmd": (
+            "gnina -r"
+            f" /inputs/{protein} -l /inputs/{ligand} -o"
+            f" /inputs/{output}"
+            f" --autobox_ligand /inputs/{protein} --cnn_scoring {cnn_scoring} --exhaustiveness 64"
+            f" --{modifier}"
+        ),
+    }
+    return json.dumps(instructions)
 
 if __name__ == "__main__":
-    print(generate_diffdock_instructions())
+    print(generate_vina_instructions())

demo.py

@@ -0,0 +1,39 @@
+import client
+import process
+import os
+import json
+
+if __name__ == "__main__":
+    casf_list = os.listdir("/home/ubuntu/casf-2016")
+    # print("generating docking instructions")
+    # docking_instruct = client.generate_diffdock_instructions()
+    # print("running docking")
+    # process.main(docking_instruct)
+    instruction_list = []
+    print("generating scoring instructions for vina")
+    for i in casf_list:
+        protein = i + '/' + i + '_protein.pdb'
+        ligand = i + '/' + i + '_ligand.sdf'
+        output = i + '/' + i + '_scored_vina.sdf.gz'
+        instruction = client.generate_vina_instructions(
+            protein=protein,
+            ligand=ligand,
+            output=output
+        )
+        print(instruction)
+        instruction_list.append(instruction)
+    print("generating scoring instructions for gnina")
+    for i in casf_list:
+        protein = i + '/' + i + '_protein.pdb'
+        ligand = i + '/' + i + '_ligand.sdf'
+        output = i + '/' + i + '_scored_gnina.sdf.gz'
+        instruction = client.generate_gnina_instructions(
+            protein=protein,
+            ligand=ligand,
+            output=output
+        )
+        print(instruction)
+        instruction_list.append(instruction)
+    print("running scoring")
+    for instruction in instruction_list:
+        process.main(json.loads(instruction))

demo.sh

@@ -1,8 +1,18 @@
+# run from basedir
+# pull container images for basic CASF 2016 evaluation
+docker pull ghcr.io/labdao/diffdock:main
 docker pull gnina/gnina
 docker pull ghcr.io/labdao/casf-2016-eval:main
+
+# pull data for basic CASF 2016 evaluation
+mkdir data
+cd data
+ipfs get bafybeiaqyjf65cs2slhilsrqvo3mo6ckdqnr5spplcts7svq7256hiiguy -o /home/ubuntu/casf-2016
+cd ..
 # avoided by adding some examples into git
 # aws s3 sync s3://labdao-benchmark/CASF-2016/coreset/1a30/ data
 
+
 ## interactive for debug
 # docker run -it -v /home/ubuntu/casf-2016-evaluator/data/coreset:/inputs -v /home/ubuntu/outputs:/outputs gnina/gnina bash
 # NOTRUN dock gnina
@@ -14,7 +24,7 @@ docker pull ghcr.io/labdao/casf-2016-eval:main
 
 # run scoring
 docker run \
-    -v /home/ubuntu/casf-2016-evaluator/data/coreset:/inputs \
+    -v /home/ubuntu/data/bafybeiaqyjf65cs2slhilsrqvo3mo6ckdqnr5spplcts7svq7256hiiguy:/inputs \
     -v /home/ubuntu/outputs:/outputs \
     gnina/gnina \
     gnina   --autobox_ligand /inputs/1a30/1a30_protein.pdb \
@@ -23,4 +33,6 @@ docker run \
             --score_only  \
             -r /inputs/1a30/1a30_protein.pdb \
             -l /inputs/1a30/1a30_ligand.sdf \
-            -o /outputs/scored_vina.sdf.gz
+            -o 1a30_scored_vina.sdf.gz &&  \
+            mkdir -p /outputs/1a30 && \
+            mv 1a30_scored_vina.sdf.gz /outputs/1a30/

install.sh

@@ -7,6 +7,10 @@
 set -E
 trap '[ "$?" -ne 77 ] || exit 77' ERR
 
+# https://unix.stackexchange.com/questions/48533/exit-shell-script-from-a-subshell
+set -E
+trap '[ "$?" -ne 77 ] || exit 77' ERR
+
 # Docker install directions from https://docs.docker.com/engine/install/ubuntu/
 echo "Installing Docker"
 sudo apt-get update

ipfs.sh

@@ -0,0 +1,18 @@
+#TODO #20
+wget https://dist.ipfs.tech/kubo/v0.18.0/kubo_v0.18.0_linux-amd64.tar.gz
+tar -xvzf kubo_v0.18.0_linux-amd64.tar.gz
+cd kubo
+sudo bash install.sh
+ipfs --version
+
+# post installation
+ipfs init
+ipfs cat /ipfs/QmQPeNsJPyVWPFDVHb77w8G42Fvo15z4bG2X8D2GhfbSXc/readme
+
+
+# port forwarding
+# 127.0.0.1:5001/webui 
+# name files with ipfs get QmWLy5XUKuSVJiYirLfw3xSxD42BDPZEVvCv7f2mGanUiD -o test.png
+
+# NOTRUN screen -ls
+screen -dm ipfs daemon -D

process.py

@@ -18,6 +18,7 @@ def __init__(self, msg: str) -> None:
 
 
 def validate_instructions(instructions: dict) -> None:
+    #TODO: #38 validate instructions if it is a dict
     for key in REQUIRED_INSTRUCTION_FIELDS:
         if key not in instructions:
             raise InputError(f"Missing required input field {key}")