Prompt changes and build caching (#15)

Author: ludomitch

src/fhda/Dockerfile.custom_deployment

@@ -16,7 +16,7 @@ RUN --mount=type=cache,target=/var/cache/apt \
     build-essential && \
     apt-get clean && rm -rf /var/lib/apt/lists/*
 
-RUN wget https://repo.anaconda.com/miniconda/Miniconda3-latest-Linux-x86_64.sh -O ~/miniconda.sh && \
+RUN wget https://repo.anaconda.com/miniconda/Miniconda3-py312_25.3.1-1-Linux-x86_64.sh -O ~/miniconda.sh && \
     chmod +x ~/miniconda.sh && \
     bash ~/miniconda.sh -b -p /app/miniconda && \
     rm ~/miniconda.sh && \
@@ -28,11 +28,14 @@ ENV PATH="/app/miniconda/bin:$PATH"
 ENV PYTHONPATH="/app/miniconda/lib/python3.12/site-packages:${PYTHONPATH:-}"
 
 # Install uv & mamba
-RUN pip3 install --no-cache-dir uv==0.5.21
-RUN conda install -c conda-forge mamba -y
+RUN --mount=type=cache,target=/root/.cache/pip \
+    pip3 install --no-cache-dir uv==0.5.21
+RUN --mount=type=cache,target=/app/miniconda/pkgs \
+    conda install -c conda-forge mamba -y
 
 # Install R and kernels in the crow_env environment
-RUN mamba install -c conda-forge -y \
+RUN --mount=type=cache,target=/app/miniconda/pkgs \
+    mamba install -c conda-forge -y \
     r-base=4.3.3 \
     r-recommended=4.3 \
     r-irkernel=1.3.2 \
@@ -83,10 +86,13 @@ RUN mamba install -c conda-forge -y \
     statsmodels=0.14.4 \
     umap-learn=0.5.7
 
-RUN python -m ipykernel install --user --name python3 --display-name "Python 3 (ipykernel)"
-RUN R -e 'IRkernel::installspec(name = "R", displayname = "R (4.3.3)")'
+RUN --mount=type=cache,target=/app/miniconda/pkgs \
+    python -m ipykernel install --user --name python3 --display-name "Python 3 (ipykernel)"
+RUN --mount=type=cache,target=/app/miniconda/pkgs \
+    R -e 'IRkernel::installspec(name = "R", displayname = "R (4.3.3)")'
 
-RUN mamba install -c conda-forge -c bioconda -y \
+RUN --mount=type=cache,target=/app/miniconda/pkgs \
+    mamba install -c conda-forge -c bioconda -y \
     biokit=0.5.0 \
     gseapy=1.1.4 \
     blast=2.16.0 \
@@ -109,22 +115,15 @@ RUN mamba install -c conda-forge -c bioconda -y \
     bioconductor-genomicranges=1.54.1 \
     bioconductor-summarizedexperiment=1.32.0 \
     bioconductor-apeglm=1.24.0 \
-    bioconductor-flowcore=2.14.0
+    bioconductor-flowcore=2.14.0 \
+    bioconductor-flowmeans=1.62.0
 
 ENV UV_COMPILE_BYTECODE=1
 ENV UV_LINK_MODE=copy
 
 FROM base AS builder
 
 ARG MODULE_NAME
-ARG USE_INTERNAL_DEPS
-ARG USE_GIT_FUTUREHOUSE_CLIENT
-
-
-RUN mkdir -p ~/.ssh && \
-    chmod 700 ~/.ssh && \
-    ssh-keyscan github.com >> ~/.ssh/known_hosts && \
-    printf "Host github.com\n    IdentityFile /root/.ssh/pqa_id_ed25519\n    IdentityFile /root/.ssh/aviary_id_ed25519\nHost gitlab.company.com\n    IdentityFile /root/.ssh/pqa_id_ed25519\n" > ~/.ssh/config
 
 RUN --mount=type=cache,target=/var/cache/apt \
     apt-get update -qq && \
@@ -135,36 +134,15 @@ RUN --mount=type=cache,target=/var/cache/apt \
 ENV VIRTUAL_ENV="/app/miniconda/bin"
 ENV PATH="/app/miniconda/bin:$PATH"
 
+# Now copy application code (this will change frequently)
 COPY ./${MODULE_NAME} /app/${MODULE_NAME}
-
 RUN mkdir -p /app/scripts
 COPY ./scripts/run_crow_job.py /app/scripts/
 
-RUN --mount=type=cache,target=/root/.cache/uv \
-    --mount=type=ssh \
-    --mount=type=secret,id=ssh_key,target=/root/.ssh/aviary_id_ed25519.tmp \
-    if [ "$USE_INTERNAL_DEPS" = "true" ]; then \
-        cp /root/.ssh/aviary_id_ed25519.tmp /root/.ssh/aviary_id_ed25519 && \
-        chmod 400 /root/.ssh/aviary_id_ed25519 && \
-        git clone git@github.com:Future-House/aviary-internal.git /app/aviary_internal && \
-        cd /app/aviary_internal/aviary_internal && \
-        uv pip install --system -e .; \
-    else \
-        echo 'Skipping aviary_internal install'; \
-    fi && \
-    if [ "$USE_GIT_FUTUREHOUSE_CLIENT" = "true" ]; then \
-        git clone git@github.com:Future-House/crow-ecosystem.git /app/crow-ecosystem && \
-        cd /app/crow-ecosystem/packages/futurehouse-client && \
-        uv pip install --system -e .; \
-    else \
-        uv pip install --system futurehouse-client; \
-    fi
-
+# Install application dependencies (this will only rerun when code changes)
 WORKDIR /app/${MODULE_NAME}
-RUN --mount=type=ssh \
-    --mount=type=secret,id=pqa_ssh_key,target=/root/.ssh/pqa_id_ed25519.tmp \
-    cp /root/.ssh/pqa_id_ed25519.tmp /root/.ssh/pqa_id_ed25519 && \
-    chmod 400 /root/.ssh/pqa_id_ed25519 && \
+RUN --mount=type=cache,target=/root/.cache/uv \
+    --mount=type=cache,target=/app/miniconda/pkgs \
     if [ -f "pyproject.toml" ]; then \
         uv pip install --system -e .; \
     elif [ -f "requirements.txt" ]; then \
@@ -176,11 +154,6 @@ RUN --mount=type=ssh \
 RUN find /app -type l -delete && \
     rm -rf /app/.git
 
-# Fix futurehouse_client imports by creating a symlink if we're using the git version
-RUN if [ "$USE_GIT_FUTUREHOUSE_CLIENT" = "true" ]; then \
-    ln -sf /app/crow-ecosystem/packages/futurehouse-client/futurehouse_client /app/miniconda/lib/python3.12/site-packages/futurehouse_client; \
-    fi
-
 FROM base AS runtime
 
 COPY --from=builder /app/ /app/

src/fhda/data_analysis_env.py

@@ -190,8 +190,6 @@ def from_task(
         # Overwrite the language in the kwargs with NBLanguage enum
         kwargs["language"] = language
         logger.info("Language: %s", language.name)
-        if language == NBLanguage.R:
-            task += f"\n{prompts.R_OUTPUT_RECOMMENDATION_PROMPT}"
 
         if trajectory_path.exists():
             files = list(trajectory_path.iterdir())

src/fhda/prompts.py

@@ -43,6 +43,8 @@
    ```r
    print(head(dataframe))
    ```
+
+4. Very important: always use the tidyverse package where possible.
 """
 
 

src/scripts/deploy.py

@@ -21,7 +21,7 @@
     "STAGE": "PROD",
 }
 
-CONTAINER_CONFIG = DockerContainerConfiguration(cpu="2", memory="4Gi")
+CONTAINER_CONFIG = DockerContainerConfiguration(cpu="8", memory="16Gi")
 
 frame_paths = [
     FramePath(path="info.cost", type="text"),