From 7d870a7750ee949efad91a6893a6c172c80ddcae Mon Sep 17 00:00:00 2001
From: hechth <helge.hecht@recetox.muni.cz>
Date: Mon, 3 Jun 2024 14:25:46 +0200
Subject: [PATCH 1/2] iitial commit

---
 tools/matchms/matchms_remove_spectra.xml      | 48 +++++++------------
 .../remove_spectra/require_filter.msp         |  1 +
 .../require_inchi_removed_spectra.msp         |  1 +
 3 files changed, 19 insertions(+), 31 deletions(-)

diff --git a/tools/matchms/matchms_remove_spectra.xml b/tools/matchms/matchms_remove_spectra.xml
index 32782885..c2f7b93a 100644
--- a/tools/matchms/matchms_remove_spectra.xml
+++ b/tools/matchms/matchms_remove_spectra.xml
@@ -1,4 +1,4 @@
-<tool id="matchms_remove_spectra" name="matchms remove spectra" version="@TOOL_VERSION@+galaxy0" profile="21.09">
+<tool id="matchms_remove_spectra" name="matchms remove spectra" version="@TOOL_VERSION@+galaxy1" profile="21.09">
     <description>Filters spectra based on metadata presence</description>
     
     <macros>
@@ -28,9 +28,8 @@ from matchms.filtering import require_compound_name, require_formula, require_pr
 require_retention_time, require_retention_index, require_valid_annotation
 from matchms.importing import load_from_mgf, load_from_msp
 
-#set metadata_fields = str("', '").join([str($f) for $f in $metadata_fields])
-required_metadata = "$metadata_fields"
-
+#set metadata_fields = [str($f) for $f in $metadata_fields]
+required_metadata = [x for x in $metadata_fields]
 if "$spectra.ext" == "msp":
     spectra = list(load_from_msp("${spectra}"))
 elif "$spectra.ext" == 'mgf':
@@ -40,57 +39,44 @@ else:
 
 filtered_spectra = []
 removed_spectra = []
-keep = False
 
 for spectrum in spectra:
+    keep = True
+
     if 'smiles' in required_metadata:
-        keep = is_valid_smiles(spectrum.get('smiles'))
+        keep = keep and is_valid_smiles(spectrum.get('smiles', '').strip())
     
     if 'inchi' in required_metadata:
-        keep = is_valid_inchi(spectrum.get('inchi'))
+        keep = keep and is_valid_inchi(spectrum.get('inchi', '').strip())
     
     if 'inchikey' in required_metadata:
-        keep = is_valid_inchikey(spectrum.get('inchikey'))
+        inchikey = spectrum.get('inchikey', '')
+        keep = keep and is_valid_inchikey(inchikey.strip())
 
     if 'precursor_mz' in required_metadata:
         result = require_precursor_mz(spectrum)
-        if result is not None:
-            keep = True
-        else:
-            keep = False
+        keep = keep and result is not None
     
     if 'valid_annotation' in required_metadata:
         result = require_valid_annotation(spectrum)
-        if result is not None:
-            keep = True
-        else:
-            keep = False
-    
+        keep = keep and result is not None
+
     if 'formula' in required_metadata:
         result = require_formula(spectrum)
-        if result is not None:
-            keep = True
-        else:
-            keep = False
+        keep = keep and result is not None
 
     if 'compound_name' in required_metadata:
         result = require_compound_name(spectrum)
-        if result is not None:
-            keep = True
-        else:
-            keep = False
-    
+        keep = keep and result is not None
+
     if 'retention_time' in required_metadata:
         result = require_retention_time(spectrum)
-        if result is not None:
-            keep = True
-        else:
-            keep = False
+        keep = keep and result is not None
     
     if 'retention_index' in required_metadata:
         result = require_retention_index(spectrum)
         if result is not None:
-            keep = True
+            keep = keep and result.get('retention_index') > 0
         else:
             keep = False
 
diff --git a/tools/matchms/test-data/remove_spectra/require_filter.msp b/tools/matchms/test-data/remove_spectra/require_filter.msp
index 46b4e9e0..73f327eb 100644
--- a/tools/matchms/test-data/remove_spectra/require_filter.msp
+++ b/tools/matchms/test-data/remove_spectra/require_filter.msp
@@ -4,6 +4,7 @@ COMMENT: SpectrumID: 1519953; Source: C4-1998-38-3; Class: Benzenoids; CASRN not
 PARENT_MASS: 347.930801
 PUBCHEMID: 10970124
 NOMINAL_MASS: 348
+RETENTION_INDEX: 0
 SMILES: nan
 NUM PEAKS: 3
 292.0       999.0
diff --git a/tools/matchms/test-data/remove_spectra/require_inchi_removed_spectra.msp b/tools/matchms/test-data/remove_spectra/require_inchi_removed_spectra.msp
index 5163cd76..ec4e5d1c 100644
--- a/tools/matchms/test-data/remove_spectra/require_inchi_removed_spectra.msp
+++ b/tools/matchms/test-data/remove_spectra/require_inchi_removed_spectra.msp
@@ -4,6 +4,7 @@ COMMENT: SpectrumID: 1519953; Source: C4-1998-38-3; Class: Benzenoids; CASRN not
 PARENT_MASS: 347.930801
 PUBCHEMID: 10970124
 NOMINAL_MASS: 348
+RETENTION_INDEX: 0.0
 SMILES: nan
 NUM PEAKS: 3
 292.0       999.0

From 5462c9424b2486722674eadce5243eb0bb8c8a75 Mon Sep 17 00:00:00 2001
From: hechth <helge.hecht@recetox.muni.cz>
Date: Mon, 3 Jun 2024 14:29:38 +0200
Subject: [PATCH 2/2] explicit float parsing

---
 tools/matchms/matchms_remove_spectra.xml | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/tools/matchms/matchms_remove_spectra.xml b/tools/matchms/matchms_remove_spectra.xml
index c2f7b93a..4ef55f46 100644
--- a/tools/matchms/matchms_remove_spectra.xml
+++ b/tools/matchms/matchms_remove_spectra.xml
@@ -76,7 +76,7 @@ for spectrum in spectra:
     if 'retention_index' in required_metadata:
         result = require_retention_index(spectrum)
         if result is not None:
-            keep = keep and result.get('retention_index') > 0
+            keep = keep and float(result.get('retention_index', 0)) > 0
         else:
             keep = False