From 289de1349d24f6f180a5c6598af0f354fd3587bf Mon Sep 17 00:00:00 2001
From: Gabriel Studer <gabriel.studer@unibas.ch>
Date: Mon, 5 Feb 2024 09:37:35 +0100
Subject: [PATCH] mmcif writer: avoid hardcoded monomer id for water chain
DOD is water too! This must be written in the chem_comp category.
Otherwise we're ending up with an invalid mmcif file. Instead of
just hardcoding, we now add whatever monomer id we observe in any
of the residues.
---
modules/io/src/mol/mmcif_writer.cc | 33 +++++++++++++++++++++++++++---
1 file changed, 30 insertions(+), 3 deletions(-)
diff --git a/modules/io/src/mol/mmcif_writer.cc b/modules/io/src/mol/mmcif_writer.cc
index e944bf9a2..4f00ef22a 100644
--- a/modules/io/src/mol/mmcif_writer.cc
+++ b/modules/io/src/mol/mmcif_writer.cc
@@ -503,6 +503,26 @@ namespace {
for(size_t i = 0; i < entity_infos.size(); ++i) {
if(entity_infos[i].type == "water" && type == "water") {
AddAsym(asym_chain_name, entity_infos[i]);
+ // there could be the situation that in the chain we saw before, only
+ // HOH represented water. This chain might suddenly have DOD.
+ // We add this to the MMCifWriterEntity to correctly write the chem_comp
+ // category in the end.
+ std::set<String> water_mon_ids(entity_infos[i].mon_ids.begin(),
+ entity_infos[i].mon_ids.end());
+ for(auto res: res_list) {
+ water_mon_ids.insert(res.GetName());
+ }
+ entity_infos[i].mon_ids = std::vector<String>(water_mon_ids.begin(),
+ water_mon_ids.end());
+ // seq and seq_can are irrelevant for water, still keep it in sync
+ std::vector<String> seq;
+ std::vector<String> seq_can;
+ for(auto mon_id: entity_infos[i].mon_ids) {
+ seq.push_back(MonIDToOLC(mon_id));
+ seq_can.push_back("?"); // It's irrelevant anyways
+ }
+ entity_infos[i].seq_olcs = seq;
+ entity_infos[i].seq_can_olcs = seq_can;
return i;
}
if(entity_infos[i].type == type &&
@@ -571,9 +591,16 @@ namespace {
}
} else {
if(type == "water") {
- mon_ids.push_back("HOH");
- seq.push_back("(HOH)");
- seq_can.push_back("?");
+ std::set<String> water_mon_ids;
+ for(auto res: res_list) {
+ water_mon_ids.insert(res.GetName());
+ }
+ mon_ids = std::vector<String>(water_mon_ids.begin(),
+ water_mon_ids.end());
+ for(auto mon_id: mon_ids) {
+ seq.push_back(MonIDToOLC(mon_id));
+ seq_can.push_back("?"); // It's irrelevant anyways
+ }
} else {
for(auto res: res_list) {
mon_ids.push_back(res.GetName());
--
GitLab