Merge pull request #484 from zargham-ahmad/issue480

Disabled key conversion in matchms tools
RECETOX · Feb 5, 2024 · da19386 · da19386
2 parents c6bc00f + d54b869
commit da19386
Show file tree

Hide file tree

Showing 30 changed files with 524 additions and 403 deletions.
diff --git a/tools/matchms/matchms_add_key.xml b/tools/matchms/matchms_add_key.xml
@@ -1,4 +1,4 @@
-<tool id="matchms_add_key" name="matchms add key" version="@TOOL_VERSION@+galaxy0" profile="21.09">
+<tool id="matchms_add_key" name="matchms add key" version="@TOOL_VERSION@+galaxy1" profile="21.09">
     <description>Set metadata key in MSP to static value</description>
 
     <macros>
@@ -25,8 +25,13 @@
 <configfile name="matchms_python_cli">
 @init_logger@
 
+import matchms
 from matchms.importing import load_from_msp
 from matchms.exporting import save_as_msp
+
+
+matchms.Metadata.set_key_replacements({})
+
 spectra = list(load_from_msp("${spectral_library}", metadata_harmonization = "False"))
 new_spectra = []
 for spectrum in spectra:
@@ -41,7 +46,15 @@ save_as_msp(new_spectra, "${output}")
             help="Mass spectral library file to add key." />
 
         <param label="Attribute Name" name="key" type="text" value="" help="Name of the attribute which will be assigned to all spectra records in the MSP." />
-        <param label="Value" name="value" type="text" value="" help="Value of the attribute which will be assigned to all spectra records in the MSP." />
+        <param label="Value" name="value" type="text" value="" help="Value of the attribute which will be assigned to all spectra records in the MSP." >
+            <sanitizer>
+                <valid initial="default">
+                    <add value="{}" />
+                    <add value="[]" />
+                    <add value="\" />
+                </valid>
+            </sanitizer>
+        </param>
     </inputs>
 
     <outputs>
@@ -54,7 +67,13 @@ save_as_msp(new_spectra, "${output}")
             <param name="spectral_library" value="filtering/input.msp" ftype="msp"/>
             <param name="key" value="tool_used"/>
             <param name="value" value="matchms"/>
-            <output name="output" file="out_matchms_add_key.msp" ftype="msp"/>
+            <output name="output" file="add_key/out_matchms_add_key.msp" ftype="msp"/>
+        </test>
+        <test>
+            <param name="spectral_library" value="add_key/add_key_test2.msp" ftype="msp"/>
+            <param name="key" value="adduct"/>
+            <param name="value" value="[M]+"/>
+            <output name="output" file="add_key/add_key_test2_out.msp" ftype="msp"/>
         </test>
     </tests>
 

diff --git a/tools/matchms/matchms_fingerprint_similarity.xml b/tools/matchms/matchms_fingerprint_similarity.xml
@@ -86,11 +86,11 @@ scores.to_json("$scores_out")
             Similarity between molecular fingerprints can serve as a proxy for structural similarity and can therefore be used to compare molecules.
 
         .. rubric:: **Footnotes**
-        .. [1] SQL join types explained on LearnSQL_.
+        .. [1] SQL join types explained on W3School_.
         .. [2] Fingerprint - the `daylight fingerprint`_ is used to compute chemical similarity.
                Fingerprints are derived from SMILES or InChI structure notations present in the spectrum metadata.
 
-        .. _LearnSQL: https://learnsql.com/blog/sql-joins-types-explained/
+        .. _W3School: https://www.w3schools.com/sql/sql_join.asp
         .. _daylight fingerprint: https://www.daylight.com/dayhtml/doc/theory/theory.finger.html
 
         @HELP_matchms@

diff --git a/tools/matchms/matchms_metadata_export.xml b/tools/matchms/matchms_metadata_export.xml
@@ -1,4 +1,4 @@
-<tool id="matchms_metadata_export" name="matchms metadata export" version="@TOOL_VERSION@+galaxy0" profile="21.09">
+<tool id="matchms_metadata_export" name="matchms metadata export" version="@TOOL_VERSION@+galaxy1" profile="21.09">
     <description>extract all metadata from mass spectra file to tabular format</description>
     <macros>
         <import>macros.xml</import>
@@ -17,9 +17,14 @@
 
 <configfiles>
 <configfile name="matchms_python_cli">
+import matchms
 from matchms.importing import load_from_msp, load_from_mgf
 from matchms.exporting.metadata_export import export_metadata_as_csv
 
+
+if "$harmonize_metadata" == "False":
+    matchms.Metadata.set_key_replacements({})
+
 spectra_list = list(load_from_${input_file.ext}("${input_file}", $harmonize_metadata))
 
 export_metadata_as_csv(spectra_list, "${output_file}")
@@ -39,10 +44,12 @@ export_metadata_as_csv(spectra_list, "${output_file}")
     <tests>
         <test>
             <param name="input_file" value="convert/mgf_out.mgf" ftype="mgf"/>
+            <param name="harmonize_metadata" value="True"/>
             <output name="output_file" file="convert/metadata.csv" ftype="csv" compare="sim_size" delta="0"/>
         </test>
         <test>
             <param name="input_file" value="similarity/RECETOX_Exposome_pesticides_HR_MS_20220323.msp" ftype="msp"/>
+            <param name="harmonize_metadata" value="True"/>
             <output name="output_file" file="convert/metadata.csv" ftype="csv" compare="sim_size" delta="0"/>
         </test>
     </tests>

diff --git a/tools/matchms/matchms_metadata_match.xml b/tools/matchms/matchms_metadata_match.xml
@@ -131,9 +131,9 @@ scores.to_json("$scores_out")
             The output will be __TRUE__ if the value is within the tolerance or an exact match and __FALSE__ otherwise.
 
         .. rubric:: **Footnotes**
-        .. [1] SQL join types explained on LearnSQL_.
+        .. [1] SQL join types explained on W3School_.
 
-        .. _LearnSQL: https://learnsql.com/blog/sql-joins-types-explained/
+        .. _W3School: https://www.w3schools.com/sql/sql_join.asp
 
         @HELP_matchms@
     </help>

diff --git a/tools/matchms/matchms_spectral_similarity.xml b/tools/matchms/matchms_spectral_similarity.xml
@@ -94,9 +94,9 @@ scores.to_json("$similarity_scores")
             For more details see this `galaxy training`_.
 
         .. rubric:: **Footnotes**
-        .. [1] SQL join types explained on LearnSQL_.
+        .. [1] SQL join types explained on W3School_.
 
-        .. _LearnSQL: https://learnsql.com/blog/sql-joins-types-explained/
+        .. _W3School: https://www.w3schools.com/sql/sql_join.asp
         .. _galaxy training: https://training.galaxyproject.org/training-material/topics/metabolomics/tutorials/gc_ms_with_xcms/tutorial.html
 
         @HELP_matchms@

diff --git a/tools/matchms/matchms_split.py b/tools/matchms/matchms_split.py
@@ -1,22 +1,13 @@
 import argparse
 import itertools
 import os
-from typing import List
 
+import matchms
 from matchms.exporting import save_as_msp
 from matchms.importing import load_from_msp
 
 
-def get_spectra_names(spectra: list) -> List[str]:
-    """Read the keyword 'compound_name' from a spectra.
-
-    Args:
-        spectra (list): List of individual spectra.
-
-    Returns:
-        List[str]: List with 'compoud_name' of individual spectra.
-    """
-    return [x.get("compound_name") for x in spectra]
+matchms.Metadata.set_key_replacements({})
 
 
 def make_outdir(outdir: str):
@@ -35,23 +26,8 @@ def write_spectra(spectra, outdir):
         spectra (List[Spectrum]): Spectra to write to file
         outdir   (str): Path to destination directory.
     """
-    names = get_spectra_names(spectra)
     for i in range(len(spectra)):
-        outpath = assemble_outpath(names[i], outdir)
-        save_as_msp(spectra[i], outpath)
-
-
-def assemble_outpath(name, outdir):
-    """Filter special chracteres from name.
-
-    Args:
-        name   (str): Name to be filetered.
-        outdir (str): Path to destination directory.
-    """
-    filename = ''.join(filter(str.isalnum, name))
-    outfile = str(filename) + ".msp"
-    outpath = os.path.join(outdir, outfile)
-    return outpath
+        save_as_msp(spectra[i], os.path.join(outdir, f"{i}.msp"))
 
 
 def split_round_robin(iterable, num_chunks):
@@ -76,7 +52,7 @@ def split_round_robin(iterable, num_chunks):
 
 
 if __name__ == "__main__":
-    spectra = load_from_msp(filename, metadata_harmonization=True)
+    spectra = load_from_msp(filename, metadata_harmonization=False)
     make_outdir(outdir)
 
     if method == "one-per-file":

diff --git a/tools/matchms/matchms_split.xml b/tools/matchms/matchms_split.xml
@@ -1,4 +1,4 @@
-<tool id="matchms_split" name="matchms split library" version="@TOOL_VERSION@+galaxy0" profile="21.09">
+<tool id="matchms_split" name="matchms split library" version="@TOOL_VERSION@+galaxy1" profile="21.09">
     <description>split a large library into subsets</description>
     <macros>
         <import>macros.xml</import>
@@ -53,16 +53,16 @@
             <param name="msp_input" value="split/sample_input.msp"  />
             <param name="split_type" value="one-per-file" />
             <output_collection name="sample" type="list">
-                <element name="1NITROPYRENE"         file="split/one-per-file/1NITROPYRENE.msp"         ftype="msp" compare="diff"/>
-                <element name="23DICHLOROPHENOL"     file="split/one-per-file/23DICHLOROPHENOL.msp"     ftype="msp" compare="diff"/>
-                <element name="245TRICHLOROPHENOL"   file="split/one-per-file/245TRICHLOROPHENOL.msp"   ftype="msp" compare="diff"/>
-                <element name="246TRICHLOROPHENOL"   file="split/one-per-file/246TRICHLOROPHENOL.msp"   ftype="msp" compare="diff"/>
-                <element name="24DICHLOROPHENOL"     file="split/one-per-file/24DICHLOROPHENOL.msp"     ftype="msp" compare="diff"/>
-                <element name="24DINITROPHENOL"      file="split/one-per-file/24DINITROPHENOL.msp"      ftype="msp" compare="diff"/>
-                <element name="25DICHLOROPHENOL"     file="split/one-per-file/25DICHLOROPHENOL.msp"     ftype="msp" compare="diff"/>
-                <element name="26DICHLOROPHENOL"     file="split/one-per-file/26DICHLOROPHENOL.msp"     ftype="msp" compare="diff"/>
-                <element name="34DICHLOROPHENOL"     file="split/one-per-file/34DICHLOROPHENOL.msp"     ftype="msp" compare="diff"/>
-                <element name="35DICHLOROPHENOL"     file="split/one-per-file/35DICHLOROPHENOL.msp"     ftype="msp" compare="diff"/>
+                <element name="0"         file="split/one-per-file/0.msp"         ftype="msp" compare="diff"/>
+                <element name="1"     file="split/one-per-file/1.msp"     ftype="msp" compare="diff"/>
+                <element name="2"   file="split/one-per-file/2.msp"   ftype="msp" compare="diff"/>
+                <element name="3"   file="split/one-per-file/3.msp"   ftype="msp" compare="diff"/>
+                <element name="4"     file="split/one-per-file/4.msp"     ftype="msp" compare="diff"/>
+                <element name="5"      file="split/one-per-file/5.msp"      ftype="msp" compare="diff"/>
+                <element name="6"     file="split/one-per-file/6.msp"     ftype="msp" compare="diff"/>
+                <element name="7"     file="split/one-per-file/7.msp"     ftype="msp" compare="diff"/>
+                <element name="8"     file="split/one-per-file/8.msp"     ftype="msp" compare="diff"/>
+                <element name="9"     file="split/one-per-file/9.msp"     ftype="msp" compare="diff"/>
             </output_collection>
         </test>
         <test>

diff --git a/tools/matchms/matchms_subsetting.xml b/tools/matchms/matchms_subsetting.xml
@@ -1,4 +1,4 @@
-<tool id="matchms_subsetting" name="matchms subsetting" version="@TOOL_VERSION@+galaxy4" profile="21.09">
+<tool id="matchms_subsetting" name="matchms subsetting" version="@TOOL_VERSION@+galaxy5" profile="21.09">
     <description>Extract spectra from a library given unique metadata identifier</description>
 
     <macros>
@@ -87,13 +87,13 @@ matchms.exporting.save_as_msp(filtered_spectra.tolist(), '${output}')
 
     <tests>
         <test>
-            <param name="spectral_library" value="out_matchms_add_key.msp" ftype="msp"/>
+            <param name="spectral_library" value="filtering/input.msp" ftype="msp"/>
             <param name="mode" value="include"/>
             <param name="list_of_identifiers" value="subsetting/identifier.csv" ftype="csv"/>
             <output name="output" file="subsetting/subsetting_output.msp" ftype="msp"/>
         </test>
         <test>
-            <param name="spectral_library" value="out_matchms_add_key.msp" ftype="msp"/>
+            <param name="spectral_library" value="filtering/input.msp" ftype="msp"/>
             <param name="mode" value="exclude"/>
             <param name="list_of_identifiers" value="subsetting/identifier.csv" ftype="csv"/>
             <output name="output" file="subsetting/subsetting_output2.msp" ftype="msp"/>

diff --git a/tools/matchms/test-data/add_key/add_key_test2.msp b/tools/matchms/test-data/add_key/add_key_test2.msp
@@ -0,0 +1,46 @@
+SCANNUMBER: -1
+IONMODE: positive
+SPECTRUMTYPE: Centroid
+FORMULA: C20H12
+INCHIKEY: CSHWQDPOILHKBI-UHFFFAOYSA-N
+SMILES: C1=CC2=C3C(=C1)C1=CC=CC4=C1C(=CC=C4)C3=CC=C2
+AUTHORS: Price et al., RECETOX, Masaryk University (CZ)
+INSTRUMENT: Q Exactive GC Orbitrap GC-MS/MS
+IONIZATION: EI+
+LICENSE: CC BY-NC
+COMPOUND_NAME: Perylene
+RETENTION_TIME: None
+RETENTION_INDEX: 2886.9
+COLLISION_ENERGY: 70eV
+INSTRUMENT_TYPE: GC-EI-Orbitrap
+CHARGE: 1
+PARENT_MASS: 251.08595400000002
+NUM PEAKS: 3
+250.07765   0.3282529462971431
+252.09323   1.0
+253.09656   0.20573802940517583
+
+SCANNUMBER: -1
+IONMODE: positive
+SPECTRUMTYPE: Centroid
+FORMULA: C14H10
+INCHIKEY: YNPNZTXNASCQKK-UHFFFAOYSA-N
+SMILES: C1=CC2=C(C=C1)C1=C(C=CC=C1)C=C2
+AUTHORS: Price et al., RECETOX, Masaryk University (CZ)
+INSTRUMENT: Q Exactive GC Orbitrap GC-MS/MS
+IONIZATION: EI+
+LICENSE: CC BY-NC
+COMPOUND_NAME: Phenanthrene
+RETENTION_TIME: None
+RETENTION_INDEX: 1832.9
+COLLISION_ENERGY: 70eV
+INSTRUMENT_TYPE: GC-EI-Orbitrap
+CHARGE: 1
+PARENT_MASS: 177.070224
+NUM PEAKS: 5
+152.0619    0.1657993569424221
+176.062     0.24558560966311757
+177.06982   0.12764433529926775
+178.0775    1.0
+179.08078   0.16394988149600653
+
diff --git a/tools/matchms/test-data/add_key/add_key_test2_out.msp b/tools/matchms/test-data/add_key/add_key_test2_out.msp
@@ -0,0 +1,48 @@
+SCANNUMBER: -1
+IONMODE: positive
+SPECTRUMTYPE: Centroid
+FORMULA: C20H12
+INCHIKEY: CSHWQDPOILHKBI-UHFFFAOYSA-N
+SMILES: C1=CC2=C3C(=C1)C1=CC=CC4=C1C(=CC=C4)C3=CC=C2
+AUTHORS: Price et al., RECETOX, Masaryk University (CZ)
+INSTRUMENT: Q Exactive GC Orbitrap GC-MS/MS
+IONIZATION: EI+
+LICENSE: CC BY-NC
+COMPOUND_NAME: Perylene
+RETENTION_TIME: None
+RETENTION_INDEX: 2886.9
+COLLISION_ENERGY: 70eV
+INSTRUMENT_TYPE: GC-EI-Orbitrap
+CHARGE: 1
+PARENT_MASS: 251.08595400000002
+ADDUCT: [M]+
+NUM PEAKS: 3
+250.07765   0.3282529462971431
+252.09323   1.0
+253.09656   0.20573802940517583
+
+SCANNUMBER: -1
+IONMODE: positive
+SPECTRUMTYPE: Centroid
+FORMULA: C14H10
+INCHIKEY: YNPNZTXNASCQKK-UHFFFAOYSA-N
+SMILES: C1=CC2=C(C=C1)C1=C(C=CC=C1)C=C2
+AUTHORS: Price et al., RECETOX, Masaryk University (CZ)
+INSTRUMENT: Q Exactive GC Orbitrap GC-MS/MS
+IONIZATION: EI+
+LICENSE: CC BY-NC
+COMPOUND_NAME: Phenanthrene
+RETENTION_TIME: None
+RETENTION_INDEX: 1832.9
+COLLISION_ENERGY: 70eV
+INSTRUMENT_TYPE: GC-EI-Orbitrap
+CHARGE: 1
+PARENT_MASS: 177.070224
+ADDUCT: [M]+
+NUM PEAKS: 5
+152.0619    0.1657993569424221
+176.062     0.24558560966311757
+177.06982   0.12764433529926775
+178.0775    1.0
+179.08078   0.16394988149600653
+
diff --git a/...matchms/test-data/out_matchms_add_key.msp → ...test-data/add_key/out_matchms_add_key.msp b/...matchms/test-data/out_matchms_add_key.msp → ...test-data/add_key/out_matchms_add_key.msp
@@ -1,8 +1,8 @@
+NAME: C001
 IONMODE: Negative
+RETENTIONTIME: 38.74
+RETENTIONINDEX: -1
 SPECTRUMTYPE: Centroid
-COMPOUND_NAME: C001
-RETENTION_TIME: 38.74
-RETENTION_INDEX: -1
 TOOL_USED: matchms
 NUM PEAKS: 57
 138.9121    10186226.0
@@ -63,11 +63,11 @@ NUM PEAKS: 57
 676.6436    1982714.0
 800.4451    2792137.0
 
+NAME: C002
 IONMODE: Negative
+RETENTIONTIME: 520.25
+RETENTIONINDEX: 1234.5
 SPECTRUMTYPE: Centroid
-COMPOUND_NAME: C002
-RETENTION_TIME: 520.25
-RETENTION_INDEX: 1234.5
 TOOL_USED: matchms
 NUM PEAKS: 35
 131.1733    1971789.0
@@ -106,10 +106,10 @@ NUM PEAKS: 35
 1216.8041   4439324.0
 1217.807    3565334.0
 
+NAME: C003
 IONMODE: Negative
+RETENTIONTIME: 483.67
 SPECTRUMTYPE: Centroid
-COMPOUND_NAME: C003
-RETENTION_TIME: 483.67
 TOOL_USED: matchms
 NUM PEAKS: 26
 265.2529    11366224.0
@@ -139,10 +139,10 @@ NUM PEAKS: 26
 1071.1639   15461047.0
 1072.1671   5096642.0
 
+NAME: C004
 IONMODE: Negative
+RETENTIONTIME: 473.48
 SPECTRUMTYPE: Centroid
-COMPOUND_NAME: C004
-RETENTION_TIME: 473.48
 TOOL_USED: matchms
 NUM PEAKS: 24
 124.1405    6517662.0
@@ -170,10 +170,10 @@ NUM PEAKS: 24
 1019.6555   57647644.0
 1020.6591   12469103.0
 
+NAME: C005
 IONMODE: Negative
+RETENTIONTIME: 41.72
 SPECTRUMTYPE: Centroid
-COMPOUND_NAME: C005
-RETENTION_TIME: 41.72
 TOOL_USED: matchms
 NUM PEAKS: 20
 218.1386    14009249.0