Address code review: move imports to top of Snakefile, rename private function

Agent-Logs-Url: https://github.com/khanlab/SPIMquant/sessions/5c9691e9-6355-4c0d-85e1-aa07f5c127e3

Co-authored-by: akhanf <11492701+akhanf@users.noreply.github.com>

Author: Copilot

spimquant/workflow/Snakefile

@@ -1,5 +1,7 @@
 import os
+from itertools import combinations as _combinations, product as _product
 
+import pandas as _pd
 from zarrnii import ZarrNii
 from snakemake.utils import format
 from snakebids import bids, generate_inputs, get_wildcard_constraints, set_bids_spec
@@ -143,9 +145,6 @@ for seg in config["crop_atlas_segs"]:
         crop_atlas_segs.append(seg)
 # Validate that model/pairwise arguments are provided when using group analysis level
 # and generate pairwise contrast labels from participants.tsv at planning time
-from itertools import combinations as _combinations, product as _product
-import pandas as _pd
-
 pairwise_contrast_labels = []
 pairwise_contrast_info = {}  # label -> dict with factor, levelA, levelB, strata
 

spimquant/workflow/scripts/perform_group_stats.py

@@ -64,7 +64,7 @@ def load_segstats_with_metadata(segstats_paths, participants_df):
     return combined.merge(participants_df, on="participant_id", how="left")
 
 
-def _build_prediction_row(region_data, pairwise_factor, level, strata):
+def build_prediction_row(region_data, pairwise_factor, level, strata):
     """Build a one-row DataFrame for marginal mean prediction.
 
     Continuous variables are held at their mean; categorical variables are
@@ -168,8 +168,8 @@ def compute_contrast_for_metric(
         fitted = smf.ols(actual_formula, data=region_data).fit()
 
         # Build prediction rows for each level at the desired strata.
-        pred_df_a = _build_prediction_row(region_data, pairwise_factor, level_a, strata)
-        pred_df_b = _build_prediction_row(region_data, pairwise_factor, level_b, strata)
+        pred_df_a = build_prediction_row(region_data, pairwise_factor, level_a, strata)
+        pred_df_b = build_prediction_row(region_data, pairwise_factor, level_b, strata)
 
         # Use patsy with the model's design_info for consistent dummy encoding.
         design_info = fitted.model.data.design_info