Updated cryo-SIM processing logic

src/murfey/client/analyser.py

@@ -219,23 +219,10 @@ def _find_context(self, file_path: Path) -> bool:
         # SIM workflow checks
         # -----------------------------------------------------------------------------
         if (
-            # CryoSIM raw data files have no extension, and end with specific suffixes
-            not file_path.suffix
-            and file_path.stem.endswith(
-                (
-                    # Bright field
-                    "_BF",
-                    # Fluorescent
-                    "_BR",
-                    "_BFR",
-                    "_GR",
-                    "_GFR",
-                    "_BR_FL",
-                    "_BFR_FL",
-                    "_GR_FL",
-                    "_GFR_FL",
-                )
-            )
+            # CryoSIM raw data files have no extension
+            # Bright field files end with "_BF"
+            # Processing will be triggered on fluorescent files ending with "_FL"
+            not file_path.suffix and file_path.stem.endswith(("_BF", "_FL"))
         ):
             if (context := _get_context("SIMContext")) is None:
                 return False

src/murfey/client/contexts/sim.py

@@ -32,16 +32,8 @@ def post_transfer(
             return None
 
         # Look for raw data files
-        # These have no extensions, and end with one of the listed suffixes
-        if not transferred_file.suffix and transferred_file.stem.endswith(
-            (
-                # Only SIM raw data files ending with '_FL' should be processed
-                "_BR_FL",
-                "_BFR_FL",
-                "_GR_FL",
-                "_GFR_FL",
-            )
-        ):
+        # These have no extensions, and end with "_FL"
+        if not transferred_file.suffix and transferred_file.stem.endswith("_FL"):
             source = _get_source(transferred_file, environment)
             if source is None:
                 logger.warning(f"No source found for file {transferred_file}")

src/murfey/server/api/workflow_sim.py

@@ -1,6 +1,7 @@
 import json
 import logging
 import re
+from functools import lru_cache
 from pathlib import Path
 from typing import Any
 
@@ -24,6 +25,34 @@
 )
 
 
+COLOR_LOOKUP = {
+    452: "blue",
+    525: "green",
+    605: "red",
+    655: "far_red",
+}
+
+
+@lru_cache(maxsize=1)
+def get_otf_files(otf_dir: Path):
+    """
+    Look for OTF files in the specified directory and match them to their wavelengths
+    """
+    otf_files: dict[str, Path] = {}
+    pattern = r"(?<!\d)\d{3}(?!\d)"  # Regex match for EXACTLY 3 consecutive digits (wavelength)
+    for file in otf_dir.glob("*"):
+        if (
+            file.is_file()
+            and file.suffix.endswith((".tif", ".tiff"))
+            and "otf" in file.stem.lower()
+            and (match := re.search(pattern, file.stem))
+        ):
+            wavelength = int(match.group())
+            if color := COLOR_LOOKUP.get(wavelength):
+                otf_files[color] = file
+    return otf_files
+
+
 class SIMDataFile(BaseModel):
     file: Path
 
@@ -50,27 +79,8 @@ def request_sim_reconstruction(
         logger.error("Error querying session information from database", exc_info=True)
         return None
 
-    # Look for OTF files in the saved directory and match them to wavelengths
-    COLOR_LOOKUP = {
-        452: "blue",
-        525: "green",
-        605: "red",
-        655: "far_red",
-    }
-    otf_files: dict[str, Path] = {}
-    pattern = r"(?<!\d)\d{3}(?!\d)"  # Regex match for EXACTLY 3 consecutive digits (wavelength)
-    for file in otf_dir.glob("*"):
-        if (
-            file.is_file()
-            and file.suffix.endswith((".tif", ".tiff"))
-            and "otf" in file.stem.lower()
-            and (match := re.search(pattern, file.stem))
-        ):
-            wavelength = int(match.group())
-            if color := COLOR_LOOKUP.get(wavelength):
-                otf_files[color] = file
-
     # If 4 matches weren't found, log as an error and exit early
+    otf_files = get_otf_files(otf_dir)
     if len(otf_files) < 4:
         logger.error(
             f"One or more OTF files missing from {otf_dir}\n"

tests/client/test_analyser.py

@@ -40,10 +40,10 @@
         # Bright field
         "visit/raw/CtrlApr_G2/20260703_132856_CtrlApr_G2_F3A_BF",
         # Fluorescent
-        "visit/raw/SR002_G1/20260707_112417_SR002G1_F1F_BR",
-        "visit/raw/SR002_G1/20260707_112417_SR002G1_F1F_BFR",
-        "visit/raw/44drug_G2/20260703_114348_44drug_G2_E2DR_GR",
-        "visit/raw/44drug_G2/20260703_113142_44drug_G2_E2DR_GFR",
+        "visit/raw/SR002_G1/20260707_112417_SR002G1_F1F_B_FL",
+        "visit/raw/SR002_G1/20260707_112417_SR002G1_F1F_G_FL",
+        "visit/raw/44drug_G2/20260703_114348_44drug_G2_E2DR_R_FL",
+        "visit/raw/44drug_G2/20260703_113142_44drug_G2_E2DR_FR_FL",
         "visit/raw/SR002_G1/20260707_112417_SR002G1_F1F_BR_FL",
         "visit/raw/SR002_G1/20260707_112417_SR002G1_F1F_BFR_FL",
         "visit/raw/44drug_G2/20260703_114348_44drug_G2_E2DR_GR_FL",