From 655f3b8348e4f06bea53602ae8696220df9b5e9d Mon Sep 17 00:00:00 2001
From: "google-labs-jules[bot]"
 <161369871+google-labs-jules[bot]@users.noreply.github.com>
Date: Sun, 18 Jan 2026 20:44:19 +0000
Subject: [PATCH] feat(perf): Optimize DataFrame construction in
 map_psms_to_spectra

Optimized the construction of the mappings DataFrame in `processing.py` by:
1. Converting the pandas Series to a list before iteration to reduce overhead.
2. Explicitly specifying columns in the `pd.DataFrame` constructor to avoid schema inference.
3. Using `dtype=object` and explicit `None` defaults to ensure missing values are treated as `None` (Falsy) rather than `NaN` (Truthy), preserving existing application logic.

Verification:
- Benchmark shows comparable or improved execution time (~0.20-0.30s).
- All tests passed (`pytest`).
- Verified `None` vs `NaN` handling for non-matches.
---
 .gitignore    |  4 +++-
 processing.py | 19 ++++++++++++-------
 2 files changed, 15 insertions(+), 8 deletions(-)

diff --git a/.gitignore b/.gitignore
index 5a2dce5..c59b0c5 100644
--- a/.gitignore
+++ b/.gitignore
@@ -1,3 +1,5 @@
 .gitignore
 /venv
-/.pytest_cache
\ No newline at end of file
+/.pytest_cache
+__pycache__/
+*.pyc
diff --git a/processing.py b/processing.py
index 5acd427..901e0c7 100644
--- a/processing.py
+++ b/processing.py
@@ -100,15 +100,20 @@ def map_psms_to_spectra(spectra: List[Dict], psm_df: pd.DataFrame) -> pd.DataFra
     # Original: Multiple apply calls (4x iteration over full dataset)
 
     # Convert matched Series to list, replacing NaNs with empty dicts for DataFrame construction
-    specs_list = [x if isinstance(x, dict) else {} for x in matched_spec_series]
-    specs_df = pd.DataFrame(specs_list)
+    # Optimization: Use tolist() for faster iteration and specify columns to avoid schema inference
+    # Use a default dict with None values to ensure consistency for non-matches.
+    defaults = {'title': None, 'mz_array': None, 'intensity_array': None, 'pepmass': None}
+
+    # We iterate over the list. If it's a dict, we use it. If not (NaN from Series), use defaults.
+    # Note: If a matching dict is missing keys, they will appear as NaN in the DataFrame unless we handle them.
+    # However, load_mgf guarantees these keys exist (even if None).
+    specs_list = [x if isinstance(x, dict) else defaults for x in matched_spec_series.tolist()]
+
+    # ⚡ OPTIMIZATION: Use dtype=object to preserve None values directly (avoids coercing to NaN)
+    # This is critical because app.py logic and tests distinguish between None (falsy) and NaN (truthy)
+    specs_df = pd.DataFrame(specs_list, columns=['title', 'mz_array', 'intensity_array', 'pepmass'], dtype=object)
     specs_df.index = psm_df.index  # Align index with original DataFrame
 
-    # Ensure required columns exist (if no spectra matched or mock data missing keys)
-    for col in ['title', 'mz_array', 'intensity_array', 'pepmass']:
-        if col not in specs_df.columns:
-            specs_df[col] = None
-
     mappings = pd.DataFrame({
         'psm_index': psm_df.index,
         'sequence': psm_df['sequence'].astype(str),