From 998dabe44b775e6cf771a9517a0086e3fd88cd4b Mon Sep 17 00:00:00 2001
From: djriffle <djriffle1@gmail.com>
Date: Wed, 4 Jun 2025 18:29:22 +0000
Subject: [PATCH] added torch and integration tools to singularity sandbox

---
 benchmarking/.gitignore                |  3 +-
 benchmarking/InteractiveAgentTester.py | 44 +++++++++++++++++++++-
 benchmarking/sandbox/requirements.txt  | 52 ++++++++++++++++++--------
 3 files changed, 80 insertions(+), 19 deletions(-)

diff --git a/benchmarking/.gitignore b/benchmarking/.gitignore
index f3a16c3..d21e39e 100644
--- a/benchmarking/.gitignore
+++ b/benchmarking/.gitignore
@@ -1,4 +1,5 @@
 .env
 __pycache__/
 .DS_store
-outputs/
\ No newline at end of file
+outputs/
+*.sif
diff --git a/benchmarking/InteractiveAgentTester.py b/benchmarking/InteractiveAgentTester.py
index 3093930..d3b06b2 100644
--- a/benchmarking/InteractiveAgentTester.py
+++ b/benchmarking/InteractiveAgentTester.py
@@ -63,6 +63,14 @@
     default="docker",
 )
 
+# Ask user whether to force‑update the sandbox image/SIF
+force_refresh = (
+    Prompt.ask(
+        "Force update sandbox environment?", choices=["y", "n"], default="n"
+    ).lower()
+    == "y"
+)
+
 is_exec_mode = backend == "singularity-exec"
 
 # -----------------------------------------------------------------------------
@@ -75,11 +83,21 @@
         from benchmarking_sandbox_management import (
             SandboxManager as _BackendManager,
             CONTAINER_NAME as _SANDBOX_HANDLE,
+            IMAGE_NAME as _SANDBOX_IMAGE,  # assume this constant exists
             API_PORT_HOST as _API_PORT,
         )
     finally:
         sys.path.pop(0)
 
+    # --- optional force‑refresh logic --------------------------------------
+    if force_refresh:
+        console.print("[yellow]Forcing Docker sandbox refresh…[/yellow]")
+        # Stop & remove any running container gracefully
+        subprocess.run(["docker", "rm", "-f", _SANDBOX_HANDLE], check=False)
+        # Remove the sandbox image to ensure re‑pull/build
+        subprocess.run(["docker", "image", "rm", "-f", _SANDBOX_IMAGE], check=False)
+        console.print("[green]Docker image removed – it will be pulled/built on next start.[/green]")
+
     def COPY_CMD(src: str, dst: str):
         subprocess.run(["docker", "cp", src, dst], check=True)
 
@@ -97,6 +115,19 @@ def COPY_CMD(src: str, dst: str):
     finally:
         sys.path.pop(0)
 
+    # optional force‑refresh
+    if force_refresh:
+        console.print("[yellow]Forcing Singularity sandbox refresh…[/yellow]")
+        try:
+            sing.stop_instance()
+        except Exception:
+            pass  # ignore if not running
+        if sing.SIF_PATH.exists():
+            sing.SIF_PATH.unlink()
+            console.print(
+                f"[green]Deleted {sing.SIF_PATH.name} – it will be re‑downloaded on next start.[/green]"
+            )
+
     class _SingInstanceWrapper:
         def start_container(self):
             return sing.start_instance()
@@ -127,9 +158,19 @@ def COPY_CMD(src: str, dst: str):
     finally:
         sys.path.pop(0)
 
+    # optional force‑refresh
+    if force_refresh:
+        console.print("[yellow]Forcing Singularity sandbox refresh…[/yellow]")
+        if sing.SIF_PATH.exists():
+            sing.SIF_PATH.unlink()
+            console.print(
+                f"[green]Deleted {sing.SIF_PATH.name} – it will be re‑downloaded on next start.[/green]"
+            )
+
     SIF_PATH = sing.SIF_PATH
     SING_BIN = sing.SING_BIN
     SENTINEL = "<<<EOF>>>"
+
     class _SingExecBackend:
         """Launch one long‑lived REPL inside the SIF and stream code to it."""
 
@@ -366,7 +407,7 @@ def format_execute_response(resp: dict) -> str:
 
 
 # ====================================================================================
-# 5 · Main interactive loop
+# 5 · Main interactive loop (unchanged)
 # ====================================================================================
 
 def run_interactive(prompt: str, dataset: Path, metadata: dict, resources: List[Tuple[Path, str]]):
@@ -481,4 +522,3 @@ def main():
         main()
     except KeyboardInterrupt:
         console.print("\nInterrupted.")
-
diff --git a/benchmarking/sandbox/requirements.txt b/benchmarking/sandbox/requirements.txt
index bdcad1e..fe48e14 100644
--- a/benchmarking/sandbox/requirements.txt
+++ b/benchmarking/sandbox/requirements.txt
@@ -1,28 +1,48 @@
-ansi2html==1.8.0
-scanpy==1.10.2
-scrublet
-anndata==0.10.8
-celltypist==1.6.3
-leidenalg==0.10.2
-igraph==0.11.6
-networkx==3.2.1
-pynndescent==0.5.13
+# Core scientific stack
 numpy==1.26.4
 scipy==1.13.1
 pandas==2.2.2
 scikit-learn==1.5.1
-umap-learn==0.5.6
-statsmodels==0.14.2
 numba==0.60.0
+statsmodels==0.14.2
+h5py==3.11.0
+
+# Visualization
 matplotlib==3.9.1
 seaborn==0.13.2
-h5py==3.11.0
+
+# Single-cell & omics
+scanpy==1.10.2
+anndata==0.10.8
+scrublet                # For doublet detection
+celltypist==1.6.3       # For cell type annotation
+
+# Dimensionality reduction & clustering
+umap-learn==0.5.6
+pynndescent==0.5.13
+leidenalg==0.10.2
+igraph==0.11.6
+networkx==3.2.1
+
+# Notebook support
+jupyter
+jupyter_client
+tqdm==4.66.4
+
+# Utilities
+ansi2html==1.8.0
 openpyxl==3.1.5
 PyPDF2
-tqdm==4.66.4
 psutil==6.0.0
 defusedxml==0.7.1
 requests==2.32.3
-jupyter
-jupyter_client
-matplotlib
\ No newline at end of file
+
+# HDF5 / Zarr data formats
+zarr
+h5netcdf
+
+# Batch correction and integration (for Scanpy)
+bbknn
+scVI-tools
+scarches
+harmonypy
\ No newline at end of file