feat: start working on docker (#5)

* feat: start working on docker

* fix: paths git clone certs etc.

* fix: docker base img

* ci: add docker workflow

* fix: path

* refactor: put pipeline in separate script

* ci: run script

* ci: dont run pipeline in build

* ci: use quay

* ci: entry point script

* ci: entrypoint

* fix: forgot shebang

* ci: cleanup

* ci: use secrets

.github/workflows/ci.yml

@@ -6,77 +6,15 @@ on:
 jobs:
   build:
     runs-on: ubuntu-latest
-
-    defaults:
-      run:
-        shell: bash -el {0}
-
     steps:
-    - uses: actions/checkout@v4
-
-    - name: Clone GT4SD
-      run: |
-        git clone https://github.com/GT4SD/gt4sd-core.git
-    
-    - uses: conda-incubator/setup-miniconda@v3
-      with:
-        activate-environment: gt4sd
-        environment-file: gt4sd-core/conda_cpu_linux.yml
-        python-version: 3.8
-        auto-activate-base: false
-        use-only-tar-bz2: true
-
-    - name: Install dependencies
-      run: |
-        conda activate gt4sd
-        cd gt4sd-core 
-        pip install --no-deps .
-        pip uninstall --yes toxsmi && pip install toxsmi
-
-    - name: Load datasets
-      run: |
-        conda activate gt4sd
-        mkdir data
-        python scripts/load_data.py \
-          --uniprot P05067 \
-          --affinity_type IC50 \
-          --affinity_cutoff 10000 \
-          --output_dir data/ \
-          --train_size 0.8 \
-          --binary_labels
-
-    - name: Train toxsmi model
-      run: |
-        conda activate gt4sd
-        python scripts/train_toxsmi.py \
-          --train data/train.csv \
-          --test data/valid.csv \
-          --smi data/mols.smi \
-          --language tokenizer \
-          --params config/toxsmi_conf.json \
-          --model models \
-          --name toxsmi_model
+      - name: Checkout code
+        uses: actions/checkout@v2
 
-    - name: Generate molecules with MoLeR
-      run: |
-        conda activate gt4sd
-        head -n 5 data/mols.smi > data/good_docks.smi
-        python scripts/moler_generate_toxsmi.py \
-          --smi_path data/good_docks.smi \
-          --param_path config/moler_conf.json \
-          --output_path data/moler_filtered \
-          --predictor_path models/toxsmi_model/weights/best_F1.pt
+      - name: Set up Docker Buildx
+        uses: docker/setup-buildx-action@v1
 
-    - name: Generate more diverse molecules with Regression Transformer
-      run: |
-        conda activate gt4sd
-        python scripts/prepare_rt_data.py \
-          --smi_path data/moler_filtered/generated.csv \
-          --output_path data/moler_filtered/generated_qed.csv
-        head -n 10 data/moler_filtered/generated_qed.csv > data/moler_filtered/generated_qed_selected.csv
-        python scripts/rt_generate.py \
-          --smi_path data/moler_filtered/generated_qed_selected.csv \
-          --param_path config/rt_conf.json \
-          --output_path data/rt
-        
+      - name: Build Docker image
+        run: docker build -t my-image:latest .
 
+      - name: Run Docker container
+        run: docker run -e API_KEY=${{ secrets.RXN_API_KEY }} -e PROJ_ID=${{ secrets.RXN_PROJ_ID }} my-image:latest

Dockerfile

@@ -0,0 +1,28 @@
+# Base image containing the installed gt4sd environment
+#FROM drugilsberg/gt4sd-base:v1.4.2-cpu
+FROM quay.io/gt4sd/gt4sd-base:v1.4.2-cpu
+
+
+# Certs for git clone
+RUN apt-get update && \
+    apt-get install -y git ca-certificates && \
+    apt-get clean
+
+RUN git clone --branch add-docker https://github.com/GT4SD/molecular-design.git
+WORKDIR /workspace/molecular-design
+
+# hack: We need to use the pypi toxsmi package, not the default one
+RUN pip uninstall --yes toxsmi && pip install toxsmi && mkdir data
+
+# hack: should be done in base gt4sd
+RUN pip uninstall --yes torch-scatter torch-sparse torch-cluster torch-geometric && \
+    pip install --no-index torch-scatter -f https://pytorch-geometric.com/whl/torch-1.12.0+cpu.html && \
+    pip install --no-index torch-sparse -f https://pytorch-geometric.com/whl/torch-1.12.0+cpu.html && \ 
+    pip install --no-index torch-cluster -f https://pytorch-geometric.com/whl/torch-1.12.0+cpu.html && \
+    pip install torch-geometric -f https://pytorch-geometric.com/whl/torch-1.12.0+cpu.html
+
+RUN chmod +x example_pipeline.sh
+
+ENTRYPOINT ["./example_pipeline.sh"]
+
+#CMD ["bash"]

example_pipeline.sh

@@ -0,0 +1,60 @@
+#!/bin/bash
+
+python scripts/load_data.py \
+      --uniprot P05067 \
+      --affinity_type IC50 \
+      --affinity_cutoff 10000 \
+      --output_dir data/ \
+      --train_size 0.8 \
+      --binary_labels
+
+# Train toxsmi model
+python scripts/train_toxsmi.py \
+      --train data/train.csv \
+      --test data/valid.csv \
+      --smi data/mols.smi \
+      --language tokenizer \
+      --params config/toxsmi_conf.json \
+      --model models \
+      --name toxsmi_model
+
+# Generate molecules with MoLeR
+head -n 5 data/mols.smi > data/good_docks.smi
+python scripts/moler_generate_toxsmi.py \
+      --smi_path data/good_docks.smi \
+      --param_path config/moler_conf.json \
+      --output_path data/moler_filtered \
+      --predictor_path models/toxsmi_model/weights/best_F1.pt
+
+# Generate more diverse molecules with Regression Transformer
+python scripts/prepare_rt_data.py \
+      --smi_path data/moler_filtered/generated.csv \
+      --output_path data/moler_filtered/generated_qed.csv && \
+head -n 10 data/moler_filtered/generated_qed.csv > data/moler_filtered/generated_qed_selected.csv
+python scripts/rt_generate.py \
+      --smi_path data/moler_filtered/generated_qed_selected.csv \
+      --param_path config/rt_conf.json \
+      --output_path data/rt
+
+python scripts/inference_dataset.py -i data/rt/qed_rt_conf_generated_qed_selected/generated.csv
+
+python scripts/test_toxsmi.py \
+    --model_path models/toxsmi_model \
+    --smi_filepath generated.smi \
+    --label_filepath dummy_data.csv \
+    --checkpoint_name F1
+
+# Calculate properties
+python scripts/mol_properties.py \
+    --smi_path models/toxsmi_model/results/dummy_data_F1_results_flat.csv \
+    --output_path mol_props.csv 
+
+# Run RXN 
+pip install rxn4chemistry
+head -n 2 data/rt/qed_rt_conf_generated_qed_selected/generated.csv > selected_for_retro.csv
+python scripts/retrosynthesis.py selected_for_retro.csv \
+    --api_key $API_KEY \
+    --project_id $PROJ_ID \
+    --steps 4 \
+    --timeout 100 \
+    --name my_retrosynthesis