Galaxy |

Changeset 30:9e3c83d2afc3 (2024-07-15)

Previous changeset 29:6501d0287dcd (2024-06-04) Next changeset 31:a5a5b147da70 (2024-11-14)

Commit message:
planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/matchms commit 8db07edd3b0d2ff778036dec410027ad58365488

modified:
macros.xml
matchms_filtering_wrapper.py
test-data/fingerprints/cosine.json
test-data/fingerprints/dice.json
test-data/fingerprints/jaccard_scores.json

diff -r 6501d0287dcd -r 9e3c83d2afc3 macros.xml
--- a/macros.xml Tue Jun 04 07:13:22 2024 +0000
+++ b/macros.xml Mon Jul 15 07:56:31 2024 +0000

@@ -1,5 +1,5 @@
<macros>
-    <token name="@TOOL_VERSION@">0.25.0</token>
+    <token name="@TOOL_VERSION@">0.27.0</token>

     <xml name="creator">
         <creator>

diff -r 6501d0287dcd -r 9e3c83d2afc3 matchms_filtering_wrapper.py
--- a/matchms_filtering_wrapper.py Tue Jun 04 07:13:22 2024 +0000
+++ b/matchms_filtering_wrapper.py Mon Jul 15 07:56:31 2024 +0000

[

@@ -2,7 +2,7 @@
import sys

from matchms.exporting import save_as_mgf, save_as_msp
-from matchms.filtering import add_compound_name, add_fingerprint, add_losses, add_parent_mass, add_precursor_mz, \
+from matchms.filtering import add_compound_name, add_fingerprint, add_parent_mass, add_precursor_mz, \
     add_retention_index, add_retention_time, clean_compound_name
from matchms.filtering import default_filters, normalize_intensities, reduce_to_number_of_peaks, select_by_mz, \
     select_by_relative_intensity
@@ -16,7 +16,7 @@
     parser.add_argument("--spectra_format", type=str, required=True, help="Format of spectra file.")
     parser.add_argument("--output", type=str, required=True, help="Filtered mass spectra file.")
     parser.add_argument("-normalise_intensities", action='store_true',
-                        help="Normalize intensities of peaks (and losses) to unit height.")
+                        help="Normalize intensities of peaks to unit height.")
     parser.add_argument("-default_filters", action='store_true',
                         help="Collection of filters that are considered default and that do no require any (factory) arguments.")
     parser.add_argument("-clean_metadata", action='store_true',
@@ -62,7 +62,7 @@
             spectrum = default_filters(spectrum)

         if args.clean_metadata:
-            filters = [add_compound_name, add_precursor_mz, add_fingerprint, add_losses, add_parent_mass,
+            filters = [add_compound_name, add_precursor_mz, add_fingerprint, add_parent_mass,
                        add_retention_index, add_retention_time, clean_compound_name]
             for metadata_filter in filters:
                 spectrum = metadata_filter(spectrum)

diff -r 6501d0287dcd -r 9e3c83d2afc3 test-data/fingerprints/cosine.json
--- a/test-data/fingerprints/cosine.json Tue Jun 04 07:13:22 2024 +0000
+++ b/test-data/fingerprints/cosine.json Mon Jul 15 07:56:31 2024 +0000

[

b'@@ -1,1 +1,1 @@\n-{"__Scores__": true, "is_symmetric": false, "references": [{"scannumber": "1161", "ionmode": "positive", "spectrumtype": "Centroid", "formula": "C4H10NO3PS", "inchikey": "YASYVMFAVPKPKE-SECBINFHSA-N", "smiles": "COP(=O)(N=C(O)C)SC", "authors": "Biomarker Analytical Laboratories, RECETOX, Masaryk University (CZ)", "instrument": "LC Orbitrap Fusion Tribrid MS", "ionization": "ESI+", "license": "CC BY-NC", "num_peaks": "16", "compound_name": "Acephate", "retention_time": 1.232997, "precursor_mz": 184.0194, "adduct": "[M+H]+", "instrument_type": "LC-ESI-Orbitrap", "peaks_json": [[90.09368, 1128.0], [93.11512, 1241.0], [95.10279, 1118.0], [101.31465, 1152.0], [102.90688, 1322.0], [103.98039, 1201.0], [112.01607, 12289.0], [112.99994, 38027.0], [115.00399, 1634.0], [124.98121, 922.0], [128.97701, 9208.0], [132.57193, 1350.0], [135.84808, 1428.0], [142.99275, 16419.0], [147.94205, 1750.0], [173.5094, 2353.0]]}, {"scannumber": "2257", "ionmode": "positive", "spectrumtype": "Centroid", "formula": "C12H11NO2", "inchikey": "CVXBEEMKQHEXEN-UHFFFAOYSA-N", "smiles": "CN=C(Oc1cccc2c1cccc2)O", "authors": "Biomarker Analytical Laboratories, RECETOX, Masaryk University (CZ)", "instrument": "LC Orbitrap Fusion Tribrid MS", "ionization": "ESI+", "license": "CC BY-NC", "peak_comments": {"145.06491": "Theoretical m/z 145.064787, Mass diff 0 (0.85 ppm), SMILES OC1=CC=CC=2C=CC=CC12, Annotation [C10H8O+H]+, Rule of HR True"}, "num_peaks": "1", "compound_name": "Carbaryl", "retention_time": 5.259445, "precursor_mz": 202.0863, "adduct": "[M+H]+", "instrument_type": "LC-ESI-Orbitrap", "peaks_json": [[145.06491, 1326147.0]]}, {"scannumber": "1516", "ionmode": "positive", "spectrumtype": "Centroid", "formula": "C8H16NO5P", "inchikey": "VEENJGZXVHKXNB-UHFFFAOYSA-N", "smiles": "COP(=O)(OC(=CC(=O)N(C)C)C)OC", "authors": "Biomarker Analytical Laboratories, RECETOX, Masaryk University (CZ)", "instrument": "LC Orbitrap Fusion Tribrid MS", "ionization": "ESI+", "license": "CC BY-NC", "peak_comments": {"112.07591": "Theoretical m/z 112.075687, Mass diff 0 (1.99 ppm), SMILES O=C(C=CC)N(C)C, Annotation [C6H11NO-H]+, Rule of HR True", "127.01563": "Theoretical m/z 127.01547, Mass diff 0 (1.26 ppm), SMILES O=P(O)(OC)OC, Annotation [C2H7O4P+H]+, Rule of HR True", "193.02605": "Theoretical m/z 193.026035, Mass diff 0 (0.08 ppm), SMILES O=CC=C(OP(=O)(OC)OC)C, Annotation [C6H11O5P-H]+, Rule of HR True", "238.08437": "Theoretical m/z 238.083891, Mass diff 0 (2.01 ppm), SMILES O=C(C=C(OP(=O)(OC)OC)C)N(C)C, Annotation [C8H16NO5P+H]+, Rule of HR True"}, "num_peaks": "5", "compound_name": "Dicrotophos", "retention_time": 2.025499, "precursor_mz": 238.0844, "adduct": "[M+H]+", "instrument_type": "LC-ESI-Orbitrap", "peaks_json": [[112.074, 102027.0], [112.07591, 9070987.0], [127.01563, 3230337.0], [193.02605, 7897744.0], [238.08437, 2973124.0]]}, {"scannumber": "1865", "ionmode": "positive", "spectrumtype": "Centroid", "formula": "C5H12NO3PS2", "inchikey": "MCWXGJITAZMZEV-UHFFFAOYSA-N", "smiles": "CN=C(CSP(=S)(OC)OC)O", "authors": "Biomarker Analytical Laboratories, RECETOX, Masaryk University (CZ)", "instrument": "LC Orbitrap Fusion Tribrid MS", "ionization": "ESI+", "license": "CC BY-NC", "peak_comments": {"88.0219": "Theoretical m/z 88.021549, Mass diff 0 (3.99 ppm), SMILES SCC=NC, Annotation [C3H7NS-H]+, Rule of HR True", "124.98233": "Theoretical m/z 124.982067, Mass diff 0 (2.11 ppm), SMILES S=P(OC)OC, Annotation [C2H7O2PS-H]+, Rule of HR True", "142.99275": "Theoretical m/z 142.993177, Mass diff 0 (0 ppm), Formula C2H8O3PS", "156.95422": "Theoretical m/z 156.954136, Mass diff 0 (0.54 ppm), SMILES S=P(S)(OC)OC, Annotation [C2H7O2PS2-H]+, Rule of HR True", "170.97": "Theoretical m/z 170.969791, Mass diff 0 (1.22 ppm), SMILES S=P(OC)(OC)SC, Annotation [C3H9O2PS2-H]+, Rule of HR True", "197.98123": "Theoretical m/z 197.980686, Mass diff 0.001 (2.75 ppm), SMILES S=P(OC)SCC(O)=NC, Annotation [C4H10NO2PS2-H]+, Rule of HR True", "198.96501": "Theoretical m/z 198'..b'mula C15H7", "189.06969": "Theoretical m/z 189.070425, Mass diff 0 (0 ppm), Formula C15H9", "201.06973": "Theoretical m/z 201.070425, Mass diff 0 (0 ppm), Formula C16H9", "233.04245": "Theoretical m/z 233.042496, Mass diff -0.001 (0 ppm), Formula C16H9S", "234.04965": "Theoretical m/z 234.049775, Mass diff 0 (0.53 ppm), SMILES S1C=2C=CC=CC2C=3C=CC=4C=CC=CC4C13, Annotation [C16H10S]+, Rule of HR False"}, "num_peaks": "23", "compound_name": "Benzo[b]naphtho[2,1-d]thiophene", "retention_time": null, "retention_index": 2419.3, "precursor_mz": 234.04965, "adduct": "[M]+", "collision_energy": "70eV", "instrument_type": "GC-EI-Orbitrap", "peaks_json": [[91.05425, 49130.0], [93.52686, 61653.0], [94.53469, 76638.0], [104.01679, 126871.0], [116.01675, 112547.0], [117.02459, 99108.0], [162.04623, 54491.0], [163.05408, 136418.0], [164.06198, 56321.0], [187.05406, 273349.0], [188.06189, 200750.0], [189.06969, 453336.0], [190.07306, 77884.0], [200.06189, 135620.0], [201.06973, 74692.0], [202.07758, 425058.0], [203.08086, 72058.0], [226.07762, 63460.0], [232.03409, 769638.0], [233.04245, 382457.0], [234.04965, 4404102.0], [235.05304, 725574.0], [236.04547, 251450.0]]}, {"scannumber": "-1", "ionmode": "positive", "spectrumtype": "Centroid", "formula": "C17H12", "inchikey": "HAPOJKSPCGLOOD-UHFFFAOYSA-N", "smiles": "C1C2=CC=CC=C2C2=C1C=C1C=CC=CC1=C2", "authors": "Price et al., RECETOX, Masaryk University (CZ)", "instrument": "Q Exactive GC Orbitrap GC-MS/MS", "ionization": "EI+", "license": "CC BY-NC", "peak_comments": {"163.05414": "Theoretical m/z 163.054775, Mass diff 0 (0 ppm), Formula C13H7", "187.05412": "Theoretical m/z 187.054775, Mass diff 0 (0 ppm), Formula C15H7", "189.06975": "Theoretical m/z 189.070425, Mass diff 0 (0 ppm), Formula C15H9", "211.05426": "Theoretical m/z 211.054775, Mass diff 0 (0 ppm), Formula C17H7", "213.0699": "Theoretical m/z 213.070425, Mass diff 0 (0 ppm), Formula C17H9", "215.0855": "Theoretical m/z 215.086075, Mass diff 0 (0 ppm), Formula C17H11", "216.09326": "Theoretical m/z 216.093354, Mass diff 0 (0.44 ppm), SMILES C=1C=CC=2C=C3C(=CC2C1)C=4C=CC=CC4C3, Annotation [C17H12]+, Rule of HR False"}, "num_peaks": "21", "compound_name": "2,3-Benzofluorene", "retention_time": null, "retention_index": 2257.5, "precursor_mz": 216.09326, "adduct": "[M]+", "collision_energy": "70eV", "instrument_type": "GC-EI-Orbitrap", "peaks_json": [[93.52689, 122035.0], [94.53471, 241743.0], [95.03638, 42130.0], [105.52688, 47526.0], [106.53471, 171653.0], [107.03639, 60004.0], [107.5425, 180288.0], [108.03385, 97471.0], [163.05414, 68194.0], [187.05412, 167183.0], [188.06201, 86262.0], [189.06975, 239421.0], [190.07321, 37778.0], [211.05426, 117044.0], [212.06215, 60813.0], [213.0699, 808499.0], [214.07317, 202222.0], [215.0855, 3634570.0], [216.09326, 2535030.0], [217.09671, 444170.0], [218.09999, 37976.0]]}], "n_row": 5, "n_col": 7, "row": [0, 0, 0, 0, 0, 0, 0, 1, 1, 1, 1, 1, 1, 1, 2, 2, 2, 2, 2, 2, 2, 3, 3, 3, 3, 3, 3, 3, 4, 4, 4, 4, 4, 4, 4], "col": [0, 1, 2, 3, 4, 5, 6, 0, 1, 2, 3, 4, 5, 6, 0, 1, 2, 3, 4, 5, 6, 0, 1, 2, 3, 4, 5, 6, 0, 1, 2, 3, 4, 5, 6], "data": [[0.05634361505508423], [0.057416923344135284], [0.12135707587003708], [0.05634361505508423], [0.06510467082262039], [0.06311869621276855], [0.16121888160705566], [0.402395635843277], [0.40114665031433105], [0.27508610486984253], [0.402395635843277], [0.15498849749565125], [0.27137377858161926], [0.2569780945777893], [0.08328129351139069], [0.08486774563789368], [0.14708952605724335], [0.08328129351139069], [0.07698478549718857], [0.12056651711463928], [0.15886470675468445], [0.06928995251655579], [0.07060988247394562], [0.1193934828042984], [0.06928995251655579], [0.09912694990634918], [0.09724045544862747], [0.1472810059785843], [0.16031154990196228], [0.15209880471229553], [0.2238437831401825], [0.16031154990196228], [0.13626770675182343], [0.20007187128067017], [0.36485710740089417]], "dtype": [["FingerprintSimilarity_cosine", "<f8"]]}\n\\ No newline at end of file\n'

diff -r 6501d0287dcd -r 9e3c83d2afc3 test-data/fingerprints/dice.json
--- a/test-data/fingerprints/dice.json Tue Jun 04 07:13:22 2024 +0000
+++ b/test-data/fingerprints/dice.json Mon Jul 15 07:56:31 2024 +0000

[

b'@@ -1,1 +1,1 @@\n-{"__Scores__": true, "is_symmetric": false, "references": [{"scannumber": "1161", "ionmode": "positive", "spectrumtype": "Centroid", "formula": "C4H10NO3PS", "inchikey": "YASYVMFAVPKPKE-SECBINFHSA-N", "smiles": "COP(=O)(N=C(O)C)SC", "authors": "Biomarker Analytical Laboratories, RECETOX, Masaryk University (CZ)", "instrument": "LC Orbitrap Fusion Tribrid MS", "ionization": "ESI+", "license": "CC BY-NC", "num_peaks": "16", "compound_name": "Acephate", "retention_time": 1.232997, "precursor_mz": 184.0194, "adduct": "[M+H]+", "instrument_type": "LC-ESI-Orbitrap", "peaks_json": [[90.09368, 1128.0], [93.11512, 1241.0], [95.10279, 1118.0], [101.31465, 1152.0], [102.90688, 1322.0], [103.98039, 1201.0], [112.01607, 12289.0], [112.99994, 38027.0], [115.00399, 1634.0], [124.98121, 922.0], [128.97701, 9208.0], [132.57193, 1350.0], [135.84808, 1428.0], [142.99275, 16419.0], [147.94205, 1750.0], [173.5094, 2353.0]]}, {"scannumber": "2257", "ionmode": "positive", "spectrumtype": "Centroid", "formula": "C12H11NO2", "inchikey": "CVXBEEMKQHEXEN-UHFFFAOYSA-N", "smiles": "CN=C(Oc1cccc2c1cccc2)O", "authors": "Biomarker Analytical Laboratories, RECETOX, Masaryk University (CZ)", "instrument": "LC Orbitrap Fusion Tribrid MS", "ionization": "ESI+", "license": "CC BY-NC", "peak_comments": {"145.06491": "Theoretical m/z 145.064787, Mass diff 0 (0.85 ppm), SMILES OC1=CC=CC=2C=CC=CC12, Annotation [C10H8O+H]+, Rule of HR True"}, "num_peaks": "1", "compound_name": "Carbaryl", "retention_time": 5.259445, "precursor_mz": 202.0863, "adduct": "[M+H]+", "instrument_type": "LC-ESI-Orbitrap", "peaks_json": [[145.06491, 1326147.0]]}, {"scannumber": "1516", "ionmode": "positive", "spectrumtype": "Centroid", "formula": "C8H16NO5P", "inchikey": "VEENJGZXVHKXNB-UHFFFAOYSA-N", "smiles": "COP(=O)(OC(=CC(=O)N(C)C)C)OC", "authors": "Biomarker Analytical Laboratories, RECETOX, Masaryk University (CZ)", "instrument": "LC Orbitrap Fusion Tribrid MS", "ionization": "ESI+", "license": "CC BY-NC", "peak_comments": {"112.07591": "Theoretical m/z 112.075687, Mass diff 0 (1.99 ppm), SMILES O=C(C=CC)N(C)C, Annotation [C6H11NO-H]+, Rule of HR True", "127.01563": "Theoretical m/z 127.01547, Mass diff 0 (1.26 ppm), SMILES O=P(O)(OC)OC, Annotation [C2H7O4P+H]+, Rule of HR True", "193.02605": "Theoretical m/z 193.026035, Mass diff 0 (0.08 ppm), SMILES O=CC=C(OP(=O)(OC)OC)C, Annotation [C6H11O5P-H]+, Rule of HR True", "238.08437": "Theoretical m/z 238.083891, Mass diff 0 (2.01 ppm), SMILES O=C(C=C(OP(=O)(OC)OC)C)N(C)C, Annotation [C8H16NO5P+H]+, Rule of HR True"}, "num_peaks": "5", "compound_name": "Dicrotophos", "retention_time": 2.025499, "precursor_mz": 238.0844, "adduct": "[M+H]+", "instrument_type": "LC-ESI-Orbitrap", "peaks_json": [[112.074, 102027.0], [112.07591, 9070987.0], [127.01563, 3230337.0], [193.02605, 7897744.0], [238.08437, 2973124.0]]}, {"scannumber": "1865", "ionmode": "positive", "spectrumtype": "Centroid", "formula": "C5H12NO3PS2", "inchikey": "MCWXGJITAZMZEV-UHFFFAOYSA-N", "smiles": "CN=C(CSP(=S)(OC)OC)O", "authors": "Biomarker Analytical Laboratories, RECETOX, Masaryk University (CZ)", "instrument": "LC Orbitrap Fusion Tribrid MS", "ionization": "ESI+", "license": "CC BY-NC", "peak_comments": {"88.0219": "Theoretical m/z 88.021549, Mass diff 0 (3.99 ppm), SMILES SCC=NC, Annotation [C3H7NS-H]+, Rule of HR True", "124.98233": "Theoretical m/z 124.982067, Mass diff 0 (2.11 ppm), SMILES S=P(OC)OC, Annotation [C2H7O2PS-H]+, Rule of HR True", "142.99275": "Theoretical m/z 142.993177, Mass diff 0 (0 ppm), Formula C2H8O3PS", "156.95422": "Theoretical m/z 156.954136, Mass diff 0 (0.54 ppm), SMILES S=P(S)(OC)OC, Annotation [C2H7O2PS2-H]+, Rule of HR True", "170.97": "Theoretical m/z 170.969791, Mass diff 0 (1.22 ppm), SMILES S=P(OC)(OC)SC, Annotation [C3H9O2PS2-H]+, Rule of HR True", "197.98123": "Theoretical m/z 197.980686, Mass diff 0.001 (2.75 ppm), SMILES S=P(OC)SCC(O)=NC, Annotation [C4H10NO2PS2-H]+, Rule of HR True", "198.96501": "Theoretical m/z 198'..b'mula C15H7", "189.06969": "Theoretical m/z 189.070425, Mass diff 0 (0 ppm), Formula C15H9", "201.06973": "Theoretical m/z 201.070425, Mass diff 0 (0 ppm), Formula C16H9", "233.04245": "Theoretical m/z 233.042496, Mass diff -0.001 (0 ppm), Formula C16H9S", "234.04965": "Theoretical m/z 234.049775, Mass diff 0 (0.53 ppm), SMILES S1C=2C=CC=CC2C=3C=CC=4C=CC=CC4C13, Annotation [C16H10S]+, Rule of HR False"}, "num_peaks": "23", "compound_name": "Benzo[b]naphtho[2,1-d]thiophene", "retention_time": null, "retention_index": 2419.3, "precursor_mz": 234.04965, "adduct": "[M]+", "collision_energy": "70eV", "instrument_type": "GC-EI-Orbitrap", "peaks_json": [[91.05425, 49130.0], [93.52686, 61653.0], [94.53469, 76638.0], [104.01679, 126871.0], [116.01675, 112547.0], [117.02459, 99108.0], [162.04623, 54491.0], [163.05408, 136418.0], [164.06198, 56321.0], [187.05406, 273349.0], [188.06189, 200750.0], [189.06969, 453336.0], [190.07306, 77884.0], [200.06189, 135620.0], [201.06973, 74692.0], [202.07758, 425058.0], [203.08086, 72058.0], [226.07762, 63460.0], [232.03409, 769638.0], [233.04245, 382457.0], [234.04965, 4404102.0], [235.05304, 725574.0], [236.04547, 251450.0]]}, {"scannumber": "-1", "ionmode": "positive", "spectrumtype": "Centroid", "formula": "C17H12", "inchikey": "HAPOJKSPCGLOOD-UHFFFAOYSA-N", "smiles": "C1C2=CC=CC=C2C2=C1C=C1C=CC=CC1=C2", "authors": "Price et al., RECETOX, Masaryk University (CZ)", "instrument": "Q Exactive GC Orbitrap GC-MS/MS", "ionization": "EI+", "license": "CC BY-NC", "peak_comments": {"163.05414": "Theoretical m/z 163.054775, Mass diff 0 (0 ppm), Formula C13H7", "187.05412": "Theoretical m/z 187.054775, Mass diff 0 (0 ppm), Formula C15H7", "189.06975": "Theoretical m/z 189.070425, Mass diff 0 (0 ppm), Formula C15H9", "211.05426": "Theoretical m/z 211.054775, Mass diff 0 (0 ppm), Formula C17H7", "213.0699": "Theoretical m/z 213.070425, Mass diff 0 (0 ppm), Formula C17H9", "215.0855": "Theoretical m/z 215.086075, Mass diff 0 (0 ppm), Formula C17H11", "216.09326": "Theoretical m/z 216.093354, Mass diff 0 (0.44 ppm), SMILES C=1C=CC=2C=C3C(=CC2C1)C=4C=CC=CC4C3, Annotation [C17H12]+, Rule of HR False"}, "num_peaks": "21", "compound_name": "2,3-Benzofluorene", "retention_time": null, "retention_index": 2257.5, "precursor_mz": 216.09326, "adduct": "[M]+", "collision_energy": "70eV", "instrument_type": "GC-EI-Orbitrap", "peaks_json": [[93.52689, 122035.0], [94.53471, 241743.0], [95.03638, 42130.0], [105.52688, 47526.0], [106.53471, 171653.0], [107.03639, 60004.0], [107.5425, 180288.0], [108.03385, 97471.0], [163.05414, 68194.0], [187.05412, 167183.0], [188.06201, 86262.0], [189.06975, 239421.0], [190.07321, 37778.0], [211.05426, 117044.0], [212.06215, 60813.0], [213.0699, 808499.0], [214.07317, 202222.0], [215.0855, 3634570.0], [216.09326, 2535030.0], [217.09671, 444170.0], [218.09999, 37976.0]]}], "n_row": 5, "n_col": 7, "row": [0, 0, 0, 0, 0, 0, 0, 1, 1, 1, 1, 1, 1, 1, 2, 2, 2, 2, 2, 2, 2, 3, 3, 3, 3, 3, 3, 3, 4, 4, 4, 4, 4, 4, 4], "col": [0, 1, 2, 3, 4, 5, 6, 0, 1, 2, 3, 4, 5, 6, 0, 1, 2, 3, 4, 5, 6, 0, 1, 2, 3, 4, 5, 6, 0, 1, 2, 3, 4, 5, 6], "data": [[0.04545454680919647], [0.04580152779817581], [0.11976047605276108], [0.04545454680919647], [0.05980066582560539], [0.06310679763555527], [0.15490533411502838], [0.31081080436706543], [0.30612245202064514], [0.2739211916923523], [0.31081080436706543], [0.13813814520835876], [0.2702702581882477], [0.25122350454330444], [0.06230529770255089], [0.0626959279179573], [0.14695340394973755], [0.06230529770255089], [0.0670391097664833], [0.11940298229455948], [0.15673981606960297], [0.05761316791176796], [0.058091286569833755], [0.11666666716337204], [0.05761316791176796], [0.09285714477300644], [0.09718669950962067], [0.13928571343421936], [0.08787878602743149], [0.08206687122583389], [0.20958751440048218], [0.08787878602743149], [0.09182209521532059], [0.17326731979846954], [0.354145348072052]], "dtype": [["FingerprintSimilarity_dice", "<f8"]]}\n\\ No newline at end of file\n'

diff -r 6501d0287dcd -r 9e3c83d2afc3 test-data/fingerprints/jaccard_scores.json
--- a/test-data/fingerprints/jaccard_scores.json Tue Jun 04 07:13:22 2024 +0000
+++ b/test-data/fingerprints/jaccard_scores.json Mon Jul 15 07:56:31 2024 +0000

[

b'@@ -1,1 +1,1 @@\n-{"__Scores__": true, "is_symmetric": false, "references": [{"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", "peak_comments": null, "num_peaks": "19", "compound_name": "Phenanthrene", "retention_time": null, "retention_index": 1832.9, "precursor_mz": 178.0775, "adduct": "[M]+", "collision_energy": "70eV", "instrument_type": "GC-EI-Orbitrap", "peaks_json": [[74.01508, 137808.0], [75.02295, 278714.0], [76.03075, 608417.0], [87.02295, 304266.0], [88.03076, 497050.0], [89.03857, 441168.0], [98.01511, 150478.0], [150.04633, 868927.0], [151.05415, 546351.0], [152.0619, 2275502.0], [153.06528, 276320.0], [169.06468, 272559.0], [174.04636, 365846.0], [175.05423, 272039.0], [176.062, 3370523.0], [177.06982, 1751846.0], [178.0775, 13724432.0], [179.08078, 2250119.0], [180.08412, 138203.0]]}, {"scannumber": "-1", "ionmode": "positive", "spectrumtype": "Centroid", "formula": "C14H10", "inchikey": "MWPLVEDNUUSJAV-UHFFFAOYSA-N", "smiles": "C1=CC2=CC3=C(C=CC=C3)C=C2C=C1", "authors": "Price et al., RECETOX, Masaryk University (CZ)", "instrument": "Q Exactive GC Orbitrap GC-MS/MS", "ionization": "EI+", "license": "CC BY-NC", "peak_comments": null, "num_peaks": "21", "compound_name": "Anthracene", "retention_time": null, "retention_index": 1844.4, "precursor_mz": 178.07754, "adduct": "[M]+", "collision_energy": "70eV", "instrument_type": "GC-EI-Orbitrap", "peaks_json": [[74.0151, 117371.0], [75.02296, 245305.0], [76.03077, 632686.0], [87.02297, 236214.0], [88.03078, 347832.0], [89.0386, 507288.0], [91.05425, 130861.0], [126.04636, 128356.0], [149.04478, 126331.0], [150.04637, 641829.0], [151.05415, 403195.0], [152.06195, 1369833.0], [153.06534, 175290.0], [169.06471, 228827.0], [174.04639, 255716.0], [175.05423, 198784.0], [176.06204, 2563053.0], [177.06984, 1182037.0], [178.07754, 11002398.0], [179.08081, 1778803.0], [180.08418, 132922.0]]}, {"scannumber": "-1", "ionmode": "positive", "spectrumtype": "Centroid", "formula": "C16H10", "inchikey": "GVEPBJHOBDJJJI-UHFFFAOYSA-N", "smiles": "C1=CC2=C(C=C1)C1=C3C2=CC=CC3=CC=C1", "authors": "Price et al., RECETOX, Masaryk University (CZ)", "instrument": "Q Exactive GC Orbitrap GC-MS/MS", "ionization": "EI+", "license": "CC BY-NC", "peak_comments": null, "num_peaks": "17", "compound_name": "Fluoranthene", "retention_time": null, "retention_index": 2102.7, "precursor_mz": 202.07756, "adduct": "[M]+", "collision_energy": "70eV", "instrument_type": "GC-EI-Orbitrap", "peaks_json": [[75.02299, 112456.0], [87.02298, 183640.0], [88.03079, 367434.0], [99.02296, 124952.0], [100.03078, 376079.0], [101.03863, 381288.0], [150.04642, 86059.0], [174.04634, 246963.0], [176.06194, 141676.0], [198.0464, 244370.0], [199.05429, 285767.0], [200.06207, 1958890.0], [201.06982, 1103710.0], [202.07756, 8104188.0], [203.08084, 1377015.0], [204.08421, 98067.0], [219.08043, 186623.0]]}, {"scannumber": "-1", "ionmode": "positive", "spectrumtype": "Centroid", "formula": "C16H10", "inchikey": "BBEAQIROQSPTKN-UHFFFAOYSA-N", "smiles": "C1=CC2=C3C(C=CC4=CC=CC(C=C2)=C34)=C1", "authors": "Price et al., RECETOX, Masaryk University (CZ)", "instrument": "Q Exactive GC Orbitrap GC-MS/MS", "ionization": "EI+", "license": "CC BY-NC", "peak_comments": null, "num_peaks": "16", "compound_name": "Pyrene", "retention_time": null, "retention_index": 2154.5, "precursor_mz": 202.07759, "adduct": "[M]+", "collision_energy": "70eV", "instrument_type": "GC-EI-Orbitrap", "peaks_json": [[87.02299, 147113.0], [88.0308, 305149.0], [99.02298, 137042.0], [100.0308, 508914.0], [101.03864, 472094.0], [101.54032, 82430.0], [150.04637, 80741.0], [174.04631, 212706.0], [198.04643, 262925.0], [199.05429, 295144.0], [200.06209, 1960712.0], [201.06982, 1270466.0], [202.'..b' [201.06982, 1270466.0], [202.07759, 7974712.0], [203.08086, 1335645.0], [204.0843, 121774.0], [219.08044, 182944.0]]}, {"scannumber": "-1", "ionmode": "positive", "spectrumtype": "Centroid", "formula": "C18H14", "inchikey": "XJKSTNDFUHDPQJ-UHFFFAOYSA-N", "smiles": "C1=CC=C(C=C1)C2=CC=C(C=C2)C3=CC=CC=C3", "authors": "Price et al., RECETOX, Masaryk University (CZ)", "instrument": "Q Exactive GC Orbitrap GC-MS/MS", "ionization": "EI+", "license": "CC BY-NC", "peak_comments": null, "num_peaks": "25", "compound_name": "para-Terphenyl", "retention_time": null, "retention_index": 2207.5, "precursor_mz": 230.10886, "adduct": "[M]+", "collision_energy": "70eV", "instrument_type": "GC-EI-Orbitrap", "peaks_json": [[88.03077, 233962.0], [101.03861, 353544.0], [113.03854, 437845.0], [115.05423, 690291.0], [128.062, 293046.0], [151.05412, 266397.0], [152.06195, 961184.0], [153.06976, 409346.0], [176.06192, 232793.0], [189.06972, 408734.0], [200.0619, 342984.0], [201.06986, 238974.0], [202.07758, 1292332.0], [203.08093, 374439.0], [213.06973, 256631.0], [215.08548, 1220866.0], [216.08881, 217320.0], [224.062, 181393.0], [226.07771, 1503774.0], [227.08588, 1198323.0], [228.09344, 2885510.0], [229.10123, 1791323.0], [230.10886, 17890056.0], [231.11214, 3536976.0], [232.11542, 320231.0]]}, {"scannumber": "-1", "ionmode": "positive", "spectrumtype": "Centroid", "formula": "C16H10S", "inchikey": "YEUHHUCOSQOCIX-UHFFFAOYSA-N", "smiles": "S1C2=C(C=CC=C2)C2=C1C1=CC=CC=C1C=C2", "authors": "Price et al., RECETOX, Masaryk University (CZ)", "instrument": "Q Exactive GC Orbitrap GC-MS/MS", "ionization": "EI+", "license": "CC BY-NC", "peak_comments": null, "num_peaks": "23", "compound_name": "Benzo[b]naphtho[2,1-d]thiophene", "retention_time": null, "retention_index": 2419.3, "precursor_mz": 234.04965, "adduct": "[M]+", "collision_energy": "70eV", "instrument_type": "GC-EI-Orbitrap", "peaks_json": [[91.05425, 49130.0], [93.52686, 61653.0], [94.53469, 76638.0], [104.01679, 126871.0], [116.01675, 112547.0], [117.02459, 99108.0], [162.04623, 54491.0], [163.05408, 136418.0], [164.06198, 56321.0], [187.05406, 273349.0], [188.06189, 200750.0], [189.06969, 453336.0], [190.07306, 77884.0], [200.06189, 135620.0], [201.06973, 74692.0], [202.07758, 425058.0], [203.08086, 72058.0], [226.07762, 63460.0], [232.03409, 769638.0], [233.04245, 382457.0], [234.04965, 4404102.0], [235.05304, 725574.0], [236.04547, 251450.0]]}, {"scannumber": "-1", "ionmode": "positive", "spectrumtype": "Centroid", "formula": "C17H12", "inchikey": "HAPOJKSPCGLOOD-UHFFFAOYSA-N", "smiles": "C1C2=CC=CC=C2C2=C1C=C1C=CC=CC1=C2", "authors": "Price et al., RECETOX, Masaryk University (CZ)", "instrument": "Q Exactive GC Orbitrap GC-MS/MS", "ionization": "EI+", "license": "CC BY-NC", "peak_comments": null, "num_peaks": "21", "compound_name": "2,3-Benzofluorene", "retention_time": null, "retention_index": 2257.5, "precursor_mz": 216.09326, "adduct": "[M]+", "collision_energy": "70eV", "instrument_type": "GC-EI-Orbitrap", "peaks_json": [[93.52689, 122035.0], [94.53471, 241743.0], [95.03638, 42130.0], [105.52688, 47526.0], [106.53471, 171653.0], [107.03639, 60004.0], [107.5425, 180288.0], [108.03385, 97471.0], [163.05414, 68194.0], [187.05412, 167183.0], [188.06201, 86262.0], [189.06975, 239421.0], [190.07321, 37778.0], [211.05426, 117044.0], [212.06215, 60813.0], [213.0699, 808499.0], [214.07317, 202222.0], [215.0855, 3634570.0], [216.09326, 2535030.0], [217.09671, 444170.0], [218.09999, 37976.0]]}], "n_row": 7, "n_col": 7, "row": [0, 0, 1, 1, 2, 3, 4, 5, 6], "col": [0, 1, 0, 1, 2, 3, 4, 5, 6], "data": [[true, true, 1.0], [true, true, 0.9629629850387573], [true, true, 0.9629629850387573], [true, true, 1.0], [true, true, 1.0], [true, true, 1.0], [true, true, 1.0], [true, true, 1.0], [true, true, 1.0]], "dtype": [["MetadataMatch_retention_index_difference_60.0_f0", "|b1"], ["MetadataMatch_retention_index_difference_20.0_f0", "|b1"], ["FingerprintSimilarity_jaccard", "<f8"]]}\n\\ No newline at end of file\n'