Merge pull request #251 from CITCOM-project/dafni-branch

Dockerised framework for DAFNI

dafni/.dockerignore

@@ -0,0 +1,4 @@
+../*
+../!causal_testing
+./!main_dafni.py
+./!data/

dafni/.env

@@ -0,0 +1,5 @@
+#.env
+VARIABLES_PATH=./data/inputs/variables.json
+CAUSAL_TESTS=./data/inputs/causal_tests.json
+DATA_PATH=./data/inputs/runtime_data.csv
+DAG_PATH=./data/inputs/dag.dot

dafni/Dockerfile

@@ -0,0 +1,33 @@
+# Define the Python version neded for CTF
+FROM python:3.10-slim
+
+## Prevents Python from writing pyc files
+ENV PYTHONDONTWRITEBYTECODE=1
+#
+## Keeps Python from buffering stdout and stderr to avoid the framework
+## from crashing without emitting any logs due to buffering
+ENV PYTHONUNBUFFERED=1
+
+#Label maintainer
+LABEL maintainer="Dr. Farhad Allian - The University of Sheffield"
+
+# Create a folder for the source code/outputs
+RUN mkdir -p ./causal_testing
+RUN mkdir -p ./data/outputs
+
+# Copy the source code and test files from build into the container
+COPY --chown=nobody  ../causal_testing ./causal_testing
+COPY --chown=nobody ./dafni/main_dafni.py ./
+COPY --chown=nobody ./dafni/data/inputs ./data/inputs
+
+# Install core dependencies using PyPi
+RUN pip install causal-testing-framework --no-cache-dir
+
+#For local testing purposes
+ENV VARIABLES_PATH=./data/inputs/variables.json \
+    CAUSAL_TESTS=./data/inputs/causal_tests.json \
+    DATA_PATH=./data/inputs/runtime_data.csv \
+    DAG_PATH=./data/inputs/dag.dot
+
+# Define the entrypoint/commands
+CMD python main_dafni.py --variables_path $VARIABLES_PATH --dag_path $DAG_PATH --data_path $DATA_PATH --tests_path $CAUSAL_TESTS

dafni/README.md

@@ -0,0 +1,29 @@
+# Causal Testing Framework on DAFNI
+
+- This directory contains the containerisation files of the causal testing framework using Docker, which is used
+to upload the framework onto [DAFNI](https://www.dafni.ac.uk).
+- It is **not** recommended to install the causal testing framework using Docker, and should only be installed
+  using [PyPI](https://pypi.org/project/causal-testing-framework/).
+
+### Folders
+
+- `data` contains two sub-folders (the structure is important for DAFNI).
+  - `inputs` is a folder that contains the input files that are (separately) uploaded to DAFNI.
+    - `causal_tests.json` is a JSON file that contains the causal tests.
+    - `variables.json` is a JSON file that contains the variables and constraints to be used.
+    - `dag.dot` is a dot file that contains the directed acyclc graph (dag) file.
+    - `runtime_data.csv` is a csv file that contains the runtime data.
+
+  - `outputs` is a folder where the `causal_tests_results.json` output file is created.
+
+### Docker files
+- `main_dafni.py` is the entry-point to the causal testing framework that is used by Docker.
+- `model_definition.yaml` is the model metadata that is required to be uploaded to DAFNI.
+- `.env` is an example of a configuration file containing the environment variables. This is only required
+    if using `docker-compose` to build the image. 
+- `Dockerfile` is the main blueprint that builds the image.
+- `.dockerignore` tells the Dockerfile which files to not include in the image.
+- `docker-compose.yaml` is another method of building the image and running the container in one line. 
+   Note: the `.env` file that contains the environment variables for `main_dafni.py` is only used here.
+
+

dafni/data/inputs/causal_tests.json

@@ -0,0 +1,136 @@
+{
+  "tests": [
+    {
+      "name": "max_doses _||_ cum_vaccinations",
+      "estimator": "LinearRegressionEstimator",
+      "estimate_type": "coefficient",
+      "effect": "direct",
+      "mutations": [
+        "max_doses"
+      ],
+      "expected_effect": {
+        "cum_vaccinations": "NoEffect"
+      },
+      "formula": "cum_vaccinations ~ max_doses",
+      "alpha": 0.05,
+      "skip": false
+    },
+    {
+      "name": "max_doses _||_ cum_vaccinated",
+      "estimator": "LinearRegressionEstimator",
+      "estimate_type": "coefficient",
+      "effect": "direct",
+      "mutations": [
+        "max_doses"
+      ],
+      "expected_effect": {
+        "cum_vaccinated": "NoEffect"
+      },
+      "formula": "cum_vaccinated ~ max_doses",
+      "alpha": 0.05,
+      "skip": false
+    },
+    {
+      "name": "max_doses _||_ cum_infections",
+      "estimator": "LinearRegressionEstimator",
+      "estimate_type": "coefficient",
+      "effect": "direct",
+      "mutations": [
+        "max_doses"
+      ],
+      "expected_effect": {
+        "cum_infections": "NoEffect"
+      },
+      "formula": "cum_infections ~ max_doses",
+      "alpha": 0.05,
+      "skip": false
+    },
+    {
+      "name": "vaccine --> cum_vaccinations",
+      "estimator": "LinearRegressionEstimator",
+      "estimate_type": "coefficient",
+      "effect": "direct",
+      "mutations": [
+        "vaccine"
+      ],
+      "expected_effect": {
+        "cum_vaccinations": "SomeEffect"
+      },
+      "formula": "cum_vaccinations ~ vaccine",
+      "skip": false
+    },
+    {
+      "name": "vaccine --> cum_vaccinated",
+      "estimator": "LinearRegressionEstimator",
+      "estimate_type": "coefficient",
+      "effect": "direct",
+      "mutations": [
+        "vaccine"
+      ],
+      "expected_effect": {
+        "cum_vaccinated": "SomeEffect"
+      },
+      "formula": "cum_vaccinated ~ vaccine",
+      "skip": false
+    },
+    {
+      "name": "vaccine --> cum_infections",
+      "estimator": "LinearRegressionEstimator",
+      "estimate_type": "coefficient",
+      "effect": "direct",
+      "mutations": [
+        "vaccine"
+      ],
+      "expected_effect": {
+        "cum_infections": "SomeEffect"
+      },
+      "formula": "cum_infections ~ vaccine",
+      "skip": false
+    },
+    {
+      "name": "cum_vaccinations _||_ cum_vaccinated | ['vaccine']",
+      "estimator": "LinearRegressionEstimator",
+      "estimate_type": "coefficient",
+      "effect": "direct",
+      "mutations": [
+        "cum_vaccinations"
+      ],
+      "expected_effect": {
+        "cum_vaccinated": "NoEffect"
+      },
+      "formula": "cum_vaccinated ~ cum_vaccinations + vaccine",
+      "alpha": 0.05,
+      "skip": false
+    },
+    {
+      "name": "cum_vaccinations _||_ cum_infections | ['vaccine']",
+      "estimator": "LinearRegressionEstimator",
+      "estimate_type": "coefficient",
+      "effect": "direct",
+      "mutations": [
+        "cum_vaccinations"
+      ],
+      "expected_effect": {
+        "cum_infections": "NoEffect"
+      },
+      "formula": "cum_infections ~ cum_vaccinations + vaccine",
+      "alpha": 0.05,
+      "skip": false
+    },
+    {
+      "name": "cum_vaccinated _||_ cum_infections | ['vaccine']",
+      "estimator": "LinearRegressionEstimator",
+      "estimate_type": "coefficient",
+      "effect": "direct",
+      "mutations": [
+        "cum_vaccinated"
+      ],
+      "expected_effect": {
+        "cum_infections": "NoEffect"
+      },
+      "formula": "cum_infections ~ cum_vaccinated + vaccine",
+      "alpha": 0.05,
+      "skip": false
+    }
+  ]
+}

dafni/data/inputs/dag.dot

@@ -0,0 +1,7 @@
+digraph CausalDAG {
+    rankdir=LR;
+    "vaccine" -> "cum_vaccinations";
+    "vaccine" -> "cum_vaccinated";
+    "vaccine" -> "cum_infections";
+    "max_doses";
+}

dafni/data/inputs/runtime_data.csv

@@ -0,0 +1,31 @@
+pop_size,pop_type,pop_infected,n_days,vaccine_type,use_waning,rand_seed,cum_infections,cum_deaths,cum_recoveries,cum_vaccinations,cum_vaccinated,target_elderly,vaccine,max_doses
+50000,hybrid,1000,50,pfizer,True,1169,6277.0,15.0,6175.0,629466.0,530715.0,True,1,2
+50000,hybrid,1000,50,pfizer,True,8888,6381.0,18.0,6274.0,630796.0,532010.0,True,1,2
+50000,hybrid,1000,50,pfizer,True,370,6738.0,15.0,6621.0,631705.0,532864.0,True,1,2
+50000,hybrid,1000,50,pfizer,True,9981,6784.0,18.0,6682.0,634582.0,535795.0,True,1,2
+50000,hybrid,1000,50,pfizer,True,6305,6757.0,20.0,6659.0,631292.0,532464.0,True,1,2
+50000,hybrid,1000,50,pfizer,True,1993,5844.0,17.0,5755.0,633314.0,534478.0,True,1,2
+50000,hybrid,1000,50,pfizer,True,1938,6465.0,19.0,6353.0,627724.0,528993.0,True,1,2
+50000,hybrid,1000,50,pfizer,True,4797,7044.0,15.0,6919.0,631246.0,532433.0,True,1,2
+50000,hybrid,1000,50,pfizer,True,2308,6878.0,6.0,6801.0,628865.0,530038.0,True,1,2
+50000,hybrid,1000,50,pfizer,True,4420,6429.0,11.0,6348.0,633803.0,535030.0,True,1,2
+50000,hybrid,1000,50,pfizer,True,2314,6566.0,15.0,6477.0,629288.0,530550.0,True,1,2
+50000,hybrid,1000,50,pfizer,True,7813,6913.0,17.0,6818.0,629290.0,530512.0,True,1,2
+50000,hybrid,1000,50,pfizer,True,1050,6963.0,14.0,6860.0,627981.0,529212.0,True,1,2
+50000,hybrid,1000,50,pfizer,True,3215,6671.0,17.0,6577.0,628802.0,530038.0,True,1,2
+50000,hybrid,1000,50,pfizer,True,2286,6597.0,13.0,6505.0,628986.0,530195.0,True,1,2
+50000,hybrid,1000,50,pfizer,True,3080,6926.0,16.0,6834.0,633636.0,534904.0,True,1,2
+50000,hybrid,1000,50,pfizer,True,7405,6438.0,15.0,6347.0,630353.0,531540.0,True,1,2
+50000,hybrid,1000,50,pfizer,True,9668,6577.0,15.0,6485.0,631257.0,532409.0,True,1,2
+50000,hybrid,1000,50,pfizer,True,8211,6197.0,13.0,6103.0,633827.0,535056.0,True,1,2
+50000,hybrid,1000,50,pfizer,True,4686,6761.0,16.0,6653.0,630557.0,531737.0,True,1,2
+50000,hybrid,1000,50,pfizer,True,3591,7328.0,24.0,7214.0,629949.0,531124.0,True,1,2
+50000,hybrid,1000,50,pfizer,True,4834,6617.0,22.0,6512.0,632609.0,533705.0,True,1,2
+50000,hybrid,1000,50,pfizer,True,6142,7017.0,17.0,6902.0,635965.0,537252.0,True,1,2
+50000,hybrid,1000,50,pfizer,True,6877,6845.0,15.0,6753.0,635678.0,536925.0,True,1,2
+50000,hybrid,1000,50,pfizer,True,1878,6480.0,20.0,6390.0,630807.0,531999.0,True,1,2
+50000,hybrid,1000,50,pfizer,True,3761,6972.0,16.0,6890.0,631100.0,532329.0,True,1,2
+50000,hybrid,1000,50,pfizer,True,1741,6581.0,20.0,6491.0,632835.0,534088.0,True,1,2
+50000,hybrid,1000,50,pfizer,True,5592,6561.0,19.0,6461.0,636799.0,537959.0,True,1,2
+50000,hybrid,1000,50,pfizer,True,7979,7075.0,17.0,6966.0,632902.0,534140.0,True,1,2
+50000,hybrid,1000,50,pfizer,True,71,6291.0,13.0,6203.0,631694.0,532901.0,True,1,2

dafni/data/inputs/variables.json

@@ -0,0 +1,47 @@
+{
+  "variables": [
+    {
+      "name": "pop_size",
+      "datatype": "int",
+      "typestring": "Input",
+      "constraint": 50000
+    },
+    {
+      "name": "pop_infected",
+      "datatype": "int",
+      "typestring": "Input",
+      "constraint": 1000
+    },
+    {
+      "name": "n_days",
+      "datatype": "int",
+      "typestring": "Input",
+      "constraint": 50
+    },
+    {
+      "name": "vaccine",
+      "datatype": "int",
+      "typestring": "Input"
+    },
+    {
+      "name": "cum_infections",
+      "datatype": "int",
+      "typestring": "Output"
+    },
+    {
+      "name": "cum_vaccinations",
+      "datatype": "int",
+      "typestring": "Output"
+    },
+    {
+      "name": "cum_vaccinated",
+      "datatype": "int",
+      "typestring": "Output"
+    },
+    {
+      "name": "max_doses",
+      "datatype": "int",
+      "typestring": "Output"
+    }
+  ]
+}