Implemented dedicated require filters matchms tool

tools/matchms/matchms_filtering.xml

@@ -1,4 +1,4 @@
-<tool id="matchms_filtering" name="matchms filtering" version="@TOOL_VERSION@+galaxy0" profile="21.09">
+<tool id="matchms_filtering" name="matchms filtering" version="@TOOL_VERSION@+galaxy1" profile="21.09">
     <description>filter and normalize mass spectrometry data</description>
 
     <macros>
@@ -49,12 +49,6 @@
                 --from_mz "$mz_range.from_mz" \
                 --to_mz "$mz_range.to_mz" \
             #end if
-            #if $require_smiles_is_true == "TRUE"
-                -require_smiles \
-            #end if
-            #if $require_inchi_is_true == "TRUE"
-                -require_inchi \
-            #end if
             #if $derive_precursor_mz_from_parent_mass.is_true == "TRUE"
                 -derive_precursor_mz_from_parent_mass \
                 --estimate_from_adduct "${derive_precursor_mz_from_parent_mass.estimate_from_adduct}" \
@@ -102,12 +96,6 @@
             <when value="FALSE"></when>
         </conditional>
 
-        <param name="require_smiles_is_true" label="Require SMILES" type="boolean" truevalue="TRUE" falsevalue="FALSE" checked="false"
-                help="Remove spectra that does not contain SMILES." />
-
-        <param name="require_inchi_is_true" label="Require INCHI" type="boolean" truevalue="TRUE" falsevalue="FALSE" checked="false"
-                help="Remove spectra that does not contain INCHI." />
-
         <conditional name="derive_precursor_mz_from_parent_mass">
             <param name="is_true" label="Derive precursor_mz from parent_mass" type="select"
                    help="Derives the precursor_mz from the parent mass and adduct or charge.">
@@ -172,16 +160,6 @@
             </section>
             <output name="output" file="filtering/mz_range.msp" ftype="msp"/>
         </test>
-        <test>
-            <param name="spectra" value="filtering/require_filter.msp" ftype="msp"/>
-            <param name="require_smiles_is_true" value="TRUE"/>
-            <output name="output" file="filtering/require_out.msp" ftype="msp"/>
-        </test>
-        <test>
-            <param name="spectra" value="filtering/require_filter.msp" ftype="msp"/>
-            <param name="require_inchi_is_true" value="TRUE"/>
-            <output name="output" file="filtering/require_out.msp" ftype="msp"/>
-        </test>
         <test>
             <param name="spectra" value="filtering/input.msp" ftype="msp"/>
             <section name="reduce_to_top_n_peaks">

tools/matchms/matchms_filtering_wrapper.py

@@ -7,18 +7,9 @@
 from matchms.filtering import default_filters, normalize_intensities, reduce_to_number_of_peaks, select_by_mz, \
     select_by_relative_intensity
 from matchms.filtering.filter_utils.derive_precursor_mz_and_parent_mass import derive_precursor_mz_from_parent_mass
-from matchms.filtering.filter_utils.smile_inchi_inchikey_conversions import is_valid_inchi, is_valid_smiles
 from matchms.importing import load_from_mgf, load_from_msp
 
 
-def require_key(spectrum, key, function):
-    value = spectrum.get(key)
-    if function(value):
-        return spectrum
-
-    return None
-
-
 def main(argv):
     parser = argparse.ArgumentParser(description="Compute MSP similarity scores")
     parser.add_argument("--spectra", type=str, required=True, help="Mass spectra file to be filtered.")
@@ -38,10 +29,6 @@ def main(argv):
                         help="Keep only peaks between set m/z range (keep if to_mz >= m/z >= from_mz).")
     parser.add_argument("--from_mz", type=float, help="Lower bound for m/z  filter")
     parser.add_argument("--to_mz", type=float, help="Upper bound for m/z  filter")
-    parser.add_argument("-require_smiles", action='store_true',
-                        help="Remove spectra that does not contain SMILES.")
-    parser.add_argument("-require_inchi", action='store_true',
-                        help="Remove spectra that does not contain INCHI.")
     parser.add_argument("-derive_precursor_mz_from_parent_mass", action='store_true',
                         help="Derives the precursor_mz from the parent mass and adduct or charge.")
     parser.add_argument("--estimate_from_adduct", type=str, help="estimate from adduct.")
@@ -55,8 +42,6 @@ def main(argv):
             or args.clean_metadata
             or args.relative_intensity
             or args.mz_range
-            or args.require_smiles
-            or args.require_inchi
             or args.derive_precursor_mz_from_parent_mass
             or args.reduce_to_top_n_peaks):
         raise ValueError('No filter selected.')
@@ -96,12 +81,6 @@ def main(argv):
             precursor_mz = derive_precursor_mz_from_parent_mass(spectrum, args.estimate_from_adduct)
             spectrum.set("precursor_mz", precursor_mz)
 
-        if args.require_smiles and spectrum is not None:
-            spectrum = require_key(spectrum, "smiles", is_valid_smiles)
-
-        if args.require_inchi and spectrum is not None:
-            spectrum = require_key(spectrum, "inchi", is_valid_inchi)
-
         if spectrum is not None:
             filtered_spectra.append(spectrum)
 

tools/matchms/matchms_remove_spectra.xml

@@ -0,0 +1,197 @@
+<tool id="matchms_remove_spectra" name="matchms remove spectra" version="@TOOL_VERSION@+galaxy0" profile="21.09">
+    <description>Filters spectra based on metadata presence</description>
+
+    <macros>
+        <import>macros.xml</import>
+    </macros>
+
+    <expand macro="creator"/>
+
+    <edam_operations>
+        <edam_operation>operation_3695</edam_operation>
+    </edam_operations>
+
+    <expand macro="bio.tools"/>
+
+    <requirements>
+        <requirement type="package" version="@TOOL_VERSION@">matchms</requirement>
+    </requirements>
+    <command detect_errors="exit_code"><![CDATA[
+        python3 '${filter_spectra}'
+    ]]></command>
+
+    <configfiles>
+        <configfile name="filter_spectra">
+from matchms.exporting import save_as_mgf, save_as_msp
+from matchms.filtering.filter_utils.smile_inchi_inchikey_conversions import is_valid_inchi, is_valid_smiles, is_valid_inchikey
+from matchms.filtering import require_compound_name, require_formula, require_precursor_mz, \
+require_retention_time, require_retention_index, require_valid_annotation
+from matchms.importing import load_from_mgf, load_from_msp
+
+#set metadata_fields = str("', '").join([str($f) for $f in $metadata_fields])
+required_metadata = "$metadata_fields"
+
+if "$spectra.ext" == "msp":
+    spectra = list(load_from_msp("${spectra}"))
+elif "$spectra.ext" == 'mgf':
+    spectra = list(load_from_mgf("${spectra}"))
+else:
+    raise ValueError(f'File format {$spectra.ext} not supported for mass spectra file.')
+
+filtered_spectra = []
+removed_spectra = []
+keep = False
+
+for spectrum in spectra:
+    if 'smiles' in required_metadata:
+        keep = is_valid_smiles(spectrum.get('smiles'))
+
+    if 'inchi' in required_metadata:
+        keep = is_valid_inchi(spectrum.get('inchi'))
+
+    if 'inchikey' in required_metadata:
+        keep = is_valid_inchikey(spectrum.get('inchikey'))
+
+    if 'precursor_mz' in required_metadata:
+        result = require_precursor_mz(spectrum)
+        if result is not None:
+            keep = True
+        else:
+            keep = False
+
+    if 'valid_annotation' in required_metadata:
+        result = require_valid_annotation(spectrum)
+        if result is not None:
+            keep = True
+        else:
+            keep = False
+
+    if 'formula' in required_metadata:
+        result = require_formula(spectrum)
+        if result is not None:
+            keep = True
+        else:
+            keep = False
+
+    if 'compound_name' in required_metadata:
+        result = require_compound_name(spectrum)
+        if result is not None:
+            keep = True
+        else:
+            keep = False
+
+    if 'retention_time' in required_metadata:
+        result = require_retention_time(spectrum)
+        if result is not None:
+            keep = True
+        else:
+            keep = False
+
+    if 'retention_index' in required_metadata:
+        result = require_retention_index(spectrum)
+        if result is not None:
+            keep = True
+        else:
+            keep = False
+
+    if keep:
+        filtered_spectra.append(spectrum)
+
+    if "$spectra_removed" == "TRUE" and keep == False:
+        removed_spectra.append(spectrum)
+
+
+if "$spectra.ext" == "msp":
+    save_as_msp(filtered_spectra, "${output_filtered}")
+    save_as_msp(removed_spectra, "${output_removed}")
+else:
+    save_as_mgf(filtered_spectra, "${output_filtered}")
+    save_as_mgf(removed_spectra, "${output_removed}")
+
+        </configfile>
+    </configfiles>
+
+    <inputs>
+        <param name="spectra" type="data" format="msp,mgf" label="Input Spectra File" help="Input file containing mass spectra"/>
+        <param name="metadata_fields" type="select" multiple="true" label="Metadata Fields" help="Select metadata fields required in the spectra">
+            <option value="smiles">SMILES</option>
+            <option value="inchi">InChI</option>
+            <option value="inchikey">InChIKey</option>
+            <option value="formula">Formula</option>
+            <option value="retention_time">Retention Time</option>
+            <option value="retention_index">Retention Index</option>
+            <option value="precursor_mz">Precursor MZ</option>
+            <option value="valid_annotation">Valid Annotation</option>
+            <option value="compound_name">Compound Name</option>
+        </param>
+        <param name="spectra_removed" type="boolean" truevalue="TRUE" falsevalue="FALSE" checked="false" label="Output Removed Spectra" help="Output spectra that were removed due to missing metadata"/>
+    </inputs>
+
+    <outputs>
+        <data name="output_filtered" format_source="spectra" label="${tool.name} on ${on_string}: Filtered Spectra"/>
+        <data name="output_removed" format_source="spectra" label="${tool.name} on ${on_string}: Removed Spectra">
+            <filter>spectra_removed</filter>
+        </data>
+    </outputs>
+
+    <tests>
+        <test expect_num_outputs="1">
+            <param name="spectra" value="remove_spectra/require_filter.msp" ftype="msp"/>
+            <param name="metadata_fields" value="smiles"/>
+            <output name="output_filtered" file="remove_spectra/require_smiles.msp" ftype="msp"/>
+        </test>
+        <test expect_num_outputs="2">
+            <param name="spectra" value="remove_spectra/require_filter.msp" ftype="msp"/>
+            <param name="metadata_fields" value="inchi"/>
+            <param name="spectra_removed" value="TRUE"/>
+            <output name="output_filtered" file="remove_spectra/require_inchi.msp" ftype="msp"/>
+            <output name="output_removed" file="remove_spectra/require_inchi_removed_spectra.msp" ftype="msp"/>
+        </test>
+        <test expect_num_outputs="1">
+            <param name="spectra" value="remove_spectra/require_filter.msp" ftype="msp"/>
+            <param name="metadata_fields" value="inchikey"/>
+            <output name="output_filtered" file="remove_spectra/require_inchikey.msp" ftype="msp"/>
+        </test>
+        <test expect_num_outputs="1">
+            <param name="spectra" value="remove_spectra/require_filter.msp" ftype="msp"/>
+            <param name="metadata_fields" value="formula"/>
+            <output name="output_filtered" file="remove_spectra/require_formula.msp" ftype="msp"/>
+        </test>
+        <test expect_num_outputs="1">
+            <param name="spectra" value="remove_spectra/require_filter.msp" ftype="msp"/>
+            <param name="metadata_fields" value="retention_time"/>
+            <output name="output_filtered" file="remove_spectra/require_retention_time.msp" ftype="msp"/>
+        </test>
+        <test expect_num_outputs="1">
+            <param name="spectra" value="remove_spectra/require_filter.msp" ftype="msp"/>
+            <param name="metadata_fields" value="retention_index"/>
+            <output name="output_filtered" file="remove_spectra/require_retention_index.msp" ftype="msp"/>
+        </test>
+        <test expect_num_outputs="1">
+            <param name="spectra" value="remove_spectra/require_filter.msp" ftype="msp"/>
+            <param name="metadata_fields" value="precursor_mz"/>
+            <output name="output_filtered" file="remove_spectra/require_precursor_mz.msp" ftype="msp"/>
+        </test>
+        <test expect_num_outputs="1">
+            <param name="spectra" value="remove_spectra/require_filter.msp" ftype="msp"/>
+            <param name="metadata_fields" value="compound_name"/>
+            <output name="output_filtered" file="remove_spectra/require_compound_name.msp" ftype="msp"/>
+        </test>
+        <test expect_num_outputs="1">
+            <param name="spectra" value="remove_spectra/require_filter.msp" ftype="msp"/>
+            <param name="metadata_fields" value="valid_annotation"/>
+            <output name="output_filtered">
+                <assert_contents>
+                    <has_n_lines n="0"/>
+                </assert_contents>
+            </output>
+        </test>
+    </tests>
+
+    <help><![CDATA[
+        This tool filters input mass spectra based on the presence of specified metadata fields. Spectra missing any of the selected metadata fields are optionally logged and output separately.
+        Valid Annotation filter removes spectra that are not fully annotated (correct and matching, smiles, inchi and inchikey)
+    ]]></help>
+
+    <expand macro="citations"/>
+</tool>