From d16c9684822bcb89d6f5268e3123c1e9cea33ce9 Mon Sep 17 00:00:00 2001
From: beale_j <john.beale@psi.ch>
Date: Wed, 11 Mar 2026 15:20:26 +0100
Subject: [PATCH] fixed chain bug

---
 generate_combined_pdb.py | 43 +++++++++++++++++++++++++++-------------
 1 file changed, 29 insertions(+), 14 deletions(-)

diff --git a/generate_combined_pdb.py b/generate_combined_pdb.py
index 0aeeb6c..9e202ad 100644
--- a/generate_combined_pdb.py
+++ b/generate_combined_pdb.py
@@ -180,24 +180,35 @@ def check_occupancies(struct):
 # -----------------------------
 def check_missing_light_residues(dark, light, ensemble):
 
-    dark_res = {(c.name, r.seqid.num, r.name) for c in dark[0] for r in c}
-    light_res = {(c.name, r.seqid.num, r.name) for c in light[0] for r in c}
-    ens_res  = {(c.name, r.seqid.num, r.name) for c in ensemble[0] for r in c}
+    ens_atoms = {
+        (c.name, r.seqid.num, r.name, a.name)
+        for c in ensemble[0]
+        for r in c
+        for a in r
+    }
 
-    missing = light_res - ens_res
+    missing = []
+
+    for chain in light[0]:
+        for res in chain:
+            for atom in res:
+
+                key = (chain.name, res.seqid.num, res.name, atom.name)
+
+                if key not in ens_atoms:
+                    missing.append(key)
 
     if missing:
 
         print()
-        print("ERROR: residues lost from light structure during merge")
-        print("------------------------------------------------------")
+        print("ERROR: atoms lost from light structure during merge")
+        print("---------------------------------------------------")
 
-        for r in sorted(missing):
-            print(" ", r)
+        for m in missing[:20]:
+            print(" ", m)
 
         print()
         print("Ensemble construction aborted.")
-        print("Check residue creation or atom filtering.")
         print()
 
         sys.exit(1)
@@ -258,11 +269,15 @@ def build_ensemble(dark, light, light_occ, occ_cutoff=0.005, verbose=False):
 
             if target_res is None:
 
-                target_res = gemmi.Residue()
-                target_res.name = res.name
-                target_res.seqid = gemmi.SeqId(res.seqid.num, res.seqid.icode)
+                new_res = gemmi.Residue()
+                new_res.name = res.name
+                new_res.seqid = gemmi.SeqId(res.seqid.num, res.seqid.icode)
+
+                target_chain.add_residue(new_res)
+
+                # IMPORTANT: retrieve the actual residue stored in the chain
+                target_res = target_chain[-1]
 
-                target_chain.add_residue(target_res)
                 res_lookup[key] = target_res
 
             for atom in res:
@@ -306,7 +321,7 @@ def build_ensemble(dark, light, light_occ, occ_cutoff=0.005, verbose=False):
 
 
 # -----------------------------
-# Mainfsortdcdds
+# Main
 # -----------------------------
 def main():