Fix retention time

app.py

@@ -264,7 +264,7 @@ def setup():
             ('https://huggingface.co/datasets/roman-bushuiev/GeMS/resolve/main/data/auxiliary/example_piper_2k_spectra.mgf',
              EXAMPLE_PATH / 'example_piper_2k_spectra.mgf',
              "PiperNET example spectra"),
-            ('https://raw.githubusercontent.com/pluskal-lab/DreaMS/cc806fa6fea281c1e57dd81fc512f71de9290017/data/examples/example_5_spectra.mgf',
+            ('https://raw.githubusercontent.com/pluskal-lab/DreaMS/refs/heads/main/data/examples/example_5_spectra.mgf',
              EXAMPLE_PATH / 'example_5_spectra.mgf',
              "DreaMS example spectra")
         ]
@@ -337,7 +337,9 @@ def _create_result_row(i, j, n, msdata, msdata_lib, sims, cos_sim, embs, calcula
     
     # Base row data
     row_data = {
-        'feature_id': i + 1,
+        'scan_number': msdata.get_values(SCAN_NUMBER, i),
+        'rt': msdata.get_values(RT, i),
+        'charge': msdata.get_values(CHARGE, i),
         'precursor_mz': msdata.get_prec_mzs(i),
         'topk': n + 1,
         'library_j': j,
@@ -377,10 +379,6 @@ def _process_results_dataframe(df, in_pth, calculate_modified_cosine=False):
     Returns:
         tuple: (processed_df, csv_path)
     """
-    # Sort hits by DreaMS similarity
-    df_top1 = df[df['topk'] == 1].sort_values('DreaMS_similarity', ascending=False)
-    df = df.set_index('feature_id').loc[df_top1['feature_id'].values].reset_index()
-    
     # Remove unnecessary columns and round similarity scores
     df = df.drop(columns=['i', 'j', 'library_j'])
     df['DreaMS_similarity'] = df['DreaMS_similarity'].astype(float).round(4)
@@ -390,12 +388,16 @@ def _process_results_dataframe(df, in_pth, calculate_modified_cosine=False):
         df['Modified_cosine_similarity'] = df['Modified_cosine_similarity'].astype(float).round(4)
     
     df['precursor_mz'] = df['precursor_mz'].astype(float).round(4)
+    df['rt'] = df['rt'].astype(float).round(2)  # Round retention time to 2 decimal places
+    df['charge'] = df['charge'].astype(str)  # Keep charge as string
     
     # Rename columns for display
     column_mapping = {
         'topk': 'Top k',
         'library_ID': 'Library ID',
-        "feature_id": "Feature ID",
+        "scan_number": "Scan number",
+        "rt": "Retention time",
+        "charge": "Charge",
         "precursor_mz": "Precursor m/z",
         "library_SMILES": "Molecule",
         "library_SMILES_raw": "SMILES",
@@ -625,11 +627,11 @@ def _create_gradio_interface():
         
         # Results table
         df = gr.Dataframe(
-            headers=["Row", "Feature ID", "Precursor m/z", "Molecule", "Spectrum", 
-                    "Library ID", "DreaMS similarity"],
-            datatype=["number", "number", "number", "html", "html", "str", "number"],
-            col_count=(7, "fixed"),
-            column_widths=["25px", "25px", "28px", "60px", "60px", "50px", "40px"],
+            headers=["Row", "Scan number", "Retention time", "Charge", "Precursor m/z", "Molecule", "Spectrum", 
+                    "DreaMS similarity", "Library ID"],
+            datatype=["number", "number", "number", "str", "number", "html", "html", "number", "str"],
+            col_count=(9, "fixed"),
+            column_widths=["20px", "30px", "30px", "25px", "30px", "40px", "40px", "40px", "50px"],
             max_height=1000,
             show_fullscreen_button=True,
             show_row_numbers=False,
@@ -643,15 +645,15 @@ def _create_gradio_interface():
         # Function to update dataframe headers based on setting
         def update_headers(show_cosine):
             if show_cosine:
-                return gr.update(headers=["Row", "Feature ID", "Precursor m/z", "Molecule", "Spectrum", 
-                                        "Library ID", "DreaMS similarity", "Modified cosine similarity"],
-                                col_count=(8, "fixed"),
-                                column_widths=["25px", "25px", "28px", "60px", "60px", "50px", "40px", "40px"])
+                return gr.update(headers=["Row", "Scan number", "Retention time", "Charge", "Precursor m/z", "Molecule", "Spectrum", 
+                                        "DreaMS similarity", "Library ID", "Modified cosine similarity"],
+                                col_count=(10, "fixed"),
+                                column_widths=["20px", "30px", "30px", "25px", "30px", "40px", "40px", "40px", "50px", "40px"])
             else:
-                return gr.update(headers=["Row", "Feature ID", "Precursor m/z", "Molecule", "Spectrum", 
-                                        "Library ID", "DreaMS similarity"],
-                                col_count=(7, "fixed"),
-                                column_widths=["25px", "25px", "28px", "60px", "60px", "50px", "40px"])
+                return gr.update(headers=["Row", "Scan number", "Retention time", "Charge", "Precursor m/z", "Molecule", "Spectrum", 
+                                        "DreaMS similarity", "Library ID"],
+                                col_count=(9, "fixed"),
+                                column_widths=["20px", "30px", "30px", "25px", "30px", "40px", "40px", "40px", "50px"])
         
         # Update headers when setting changes
         calculate_modified_cosine.change(

data/example_5_spectra.mgf

@@ -1,350 +0,0 @@
-BEGIN IONS
-NAME=DMAPT
-DESCRIPTION=MCE bioactive compounds library
-EXACTMASS=293.199094
-FORMULA=C17H27NO3
-INCHI=InChI=1S/C17H27NO3/c1-11-6-5-9-17(2)15(21-17)14-12(8-7-11)13(10-18(3)4)16(19)20-14/h6,12-15H,5,7-10H2,1-4H3/b11-6-/t12-,13+,14-,15+,17+/m0/s1
-INCHIAUX=UJNSFDHVIBGEJZ-CMRIBGNTSA-N
-SMILES=C/C1=C/CC[C@@]2(C)O[C@@H]2[C@H]2OC(=O)[C@H](CN(C)C)[C@@H]2CC1
-FEATURE_ID=-1
-MSLEVEL=2
-RTINSECONDS=69.34
-ADDUCT=[M+H]+
-PEPMASS=294.20637
-CHARGE=1
-SPECTYPE=ALL_ENERGIES
-Collision energy=60.0
-FRAGMENTATION_METHOD=HCD
-ISOLATION_WINDOW=1.2000000476839432
-Acquisition=Crude
-INSTRUMENT_TYPE=Orbitrap
-SOURCE_INSTRUMENT=Orbitrap ID-X
-IMS_TYPE=none
-ION_SOURCE=ESI
-IONMODE=Positive
-PI=Tomas Pluskal
-DATACOLLECTOR=Corinna Brungs
-DATASET_ID=MSVPLACEHOLDERID
-USI=mzspec:MSVPLACEHOLDERID:20220601_100AGC_pluskal_mce_1D1_A13_id.mzML:-1
-SCANS=-1
-PRECURSOR_PURITY=1.0
-QUALITY_CHIMERIC=PASSED
-QUALITY_EXPLAINED_INTENSITY=0.95719075
-QUALITY_EXPLAINED_SIGNALS=0.91803277
-Num peaks=61
-42.033739 2.023
-43.017743 1.244
-43.041538 0.375
-44.049385 0.271
-46.064932 0.633
-55.053921 0.247
-56.049297 0.434
-58.061086 0.921
-58.064829 100
-58.068315 4.309
-58.071661 1.067
-58.074808 0.3
-67.053963 0.832
-69.069565 0.262
-79.053978 1.014
-81.069616 1.317
-82.064804 0.901
-84.080497 0.567
-91.053903 1.516
-93.069577 2.084
-94.064888 0.298
-95.048843 0.28
-95.085147 0.577
-97.088348 0.201
-98.059745 0.372
-105.069597 1.626
-106.064804 0.358
-107.085253 1.159
-108.080338 0.242
-109.064416 0.257
-109.100923 0.465
-110.096075 1.043
-116.070213 0.429
-117.069763 0.601
-119.085279 1.489
-121.100884 0.483
-129.069626 0.285
-131.085311 0.665
-133.10089 0.474
-134.059753 0.937
-135.117294 0.205
-144.093231 0.522
-145.100601 0.202
-147.11673 0.298
-149.13218 2.214
-159.116485 2.567
-161.131911 0.35
-164.107071 0.243
-164.143127 0.255
-175.112091 0.209
-177.127193 1.388
-185.13264 0.432
-192.174698 0.38
-203.14307 0.317
-222.185104 0.206
-231.13818 0.57
-249.148476 2.331
-250.216614 0.841
-251.117976 1.62
-294.128296 0.191
-294.206632 26.983
-END IONS
-
-BEGIN IONS
-NAME=Mirk-IN-1
-DESCRIPTION=MCE bioactive compounds library
-EXACTMASS=497.065759
-FORMULA=C23H17Cl2N5O4
-INCHI=InChI=1S/C23H17Cl2N5O4/c1-34-23-27-11-14-8-16(22(33)29-19(14)30-23)21(32)28-18-9-13(5-6-17(18)25)20(31)26-10-12-3-2-4-15(24)7-12/h2-9,11H,10H2,1H3,(H,26,31)(H,28,32)(H,27,29,30,33)
-INCHIAUX=CQKBSRPVZZLCJE-UHFFFAOYSA-N
-SMILES=COc1ncc2cc(C(=O)Nc3c(Cl)ccc(C(=O)NCc4cc(Cl)ccc4)c3)c(=O)[nH]c2n1
-FEATURE_ID=-1
-MSLEVEL=2
-RTINSECONDS=115.08
-ADDUCT=[M+H]+
-PEPMASS=498.07304
-CHARGE=1
-SPECTYPE=ALL_MSN_TO_PSEUDO_MS2
-Collision energy=60.0
-FRAGMENTATION_METHOD=HCD
-ISOLATION_WINDOW=1.2000000476839432
-Acquisition=Crude
-INSTRUMENT_TYPE=Orbitrap
-SOURCE_INSTRUMENT=Orbitrap ID-X
-IMS_TYPE=none
-ION_SOURCE=ESI
-IONMODE=Positive
-PI=Tomas Pluskal
-DATACOLLECTOR=Corinna Brungs
-DATASET_ID=MSVPLACEHOLDERID
-USI=mzspec:MSVPLACEHOLDERID:20220601_pluskal_mce_1D1_A4_id.mzML:-1
-SCANS=-1
-PRECURSOR_PURITY=1.0
-QUALITY_CHIMERIC=PASSED
-QUALITY_EXPLAINED_INTENSITY=0.93082154
-QUALITY_EXPLAINED_SIGNALS=0.88461536
-Num peaks=52
-42.033703 6.933
-58.028778 11.37
-59.032055 2.116
-60.03299 1.268
-63.022655 2.006
-64.017876 2.374
-65.038353 0.923
-66.033494 2.134
-69.992175 1.938
-78.033541 4.097
-79.028745 1.257
-86.060081 9.959
-87.06353 3.769
-88.064438 1.499
-91.017532 3.789
-93.044386 7.213
-104.049591 5.292
-105.044537 4.776
-105.052979 3.494
-106.028389 1.142
-120.055412 7.778
-121.039436 6.751
-125.014962 2.948
-133.027011 2.074
-133.039425 3.979
-135.018616 0.908
-148.050389 4.721
-154.005085 2.139
-160.050642 3.311
-161.034588 1.467
-176.045663 22.882
-194.056314 24.336
-204.040476 24.707
-221.080734 1.76
-222.05116 100
-222.087845 1.473
-223.089127 1.679
-248.069031 1.01
-249.074326 2.424
-250.078537 2.556
-251.081848 1.513
-277.029205 1.552
-293.066956 1.309
-308.148193 1.723
-309.15033 3.711
-310.153687 1.904
-321.062225 1.269
-357.03932 4.785
-411.191315 1.423
-412.193907 5.738
-413.196318 8.005
-498.073273 1.812
-END IONS
-
-BEGIN IONS
-NAME=1373215-15-6
-DESCRIPTION=MCE bioactive compounds library
-EXACTMASS=484.212198
-FORMULA=C25H29FN4O5
-INCHI=InChI=1S/C25H29FN4O5/c1-27-25(32)35-22-15-34-21-7-6-18(29-8-10-30(11-9-29)19-13-33-14-19)12-20(21)23(22)28-24(31)16-2-4-17(26)5-3-16/h2-7,12,19,22-23H,8-11,13-15H2,1H3,(H,27,32)(H,28,31)/t22-,23-/m0/s1
-INCHIAUX=NDEBZCZEAVMSQF-GOTSBHOMSA-N
-SMILES=CNC(=O)O[C@H]1COc2c(cc(N3CCN(C4COC4)CC3)cc2)[C@@H]1NC(=O)c1ccc(F)cc1
-FEATURE_ID=-1
-MSLEVEL=2
-RTINSECONDS=77.81
-ADDUCT=[M+H]+
-PEPMASS=485.21947
-CHARGE=1
-SPECTYPE=SAME_ENERGY
-Collision energy=30.0
-FRAGMENTATION_METHOD=HCD
-ISOLATION_WINDOW=1.2000000476839432
-Acquisition=Crude
-INSTRUMENT_TYPE=Orbitrap
-SOURCE_INSTRUMENT=Orbitrap ID-X
-IMS_TYPE=none
-ION_SOURCE=ESI
-IONMODE=Positive
-PI=Tomas Pluskal
-DATACOLLECTOR=Corinna Brungs
-DATASET_ID=MSVPLACEHOLDERID
-USI=mzspec:MSVPLACEHOLDERID:20220601_pluskal_mce_1D1_A8_id.mzML:-1
-SCANS=-1
-PRECURSOR_PURITY=1.0
-QUALITY_CHIMERIC=PASSED
-QUALITY_EXPLAINED_INTENSITY=1.0
-QUALITY_EXPLAINED_SIGNALS=1.0
-Num peaks=15
-70.064962 2.862
-84.080627 1.511
-109.076332 1.385
-114.091324 2.886
-123.023911 14.149
-161.059692 2.52
-202.086288 1.435
-230.105148 3.367
-241.133691 26.664
-261.159847 3.062
-271.144251 41.501
-289.154956 100
-346.175893 26.1
-351.136749 1.2
-410.186693 8.75
-END IONS
-
-BEGIN IONS
-NAME=IPSU
-DESCRIPTION=MCE bioactive compounds library
-EXACTMASS=405.216475
-FORMULA=C23H27N5O2
-INCHI=InChI=1S/C23H27N5O2/c1-30-20-7-11-24-22(26-20)27-13-9-23(10-14-27)8-4-12-28(21(23)29)16-17-15-25-19-6-3-2-5-18(17)19/h2-3,5-7,11,15,25H,4,8-10,12-14,16H2,1H3
-INCHIAUX=PCMHOSYCWRRHTG-UHFFFAOYSA-N
-SMILES=COc1nc(N2CCC3(CCCN(Cc4c[nH]c5ccccc45)C3=O)CC2)ncc1
-FEATURE_ID=660
-MSLEVEL=2
-RTINSECONDS=110.45
-ADDUCT=[M+H]+
-PEPMASS=406.22375
-CHARGE=1
-Collision energy=20.0
-FRAGMENTATION_METHOD=HCD
-ISOLATION_WINDOW=1.2000000476839432
-Acquisition=Crude
-INSTRUMENT_TYPE=Orbitrap
-SOURCE_INSTRUMENT=Orbitrap ID-X
-IMS_TYPE=none
-ION_SOURCE=ESI
-IONMODE=Positive
-PI=Tomas Pluskal
-DATACOLLECTOR=Corinna Brungs
-DATASET_ID=MSVPLACEHOLDERID
-USI=mzspec:MSVPLACEHOLDERID:20220601_pluskal_mce_1D1_A1_id.mzML:660
-SCANS=660
-PRECURSOR_PURITY=0.9731724062527856
-QUALITY_CHIMERIC=PASSED
-QUALITY_EXPLAINED_INTENSITY=0.99532574
-QUALITY_EXPLAINED_SIGNALS=0.8888889
-Num peaks=9
-45.134823 1.082
-45.13699 1.064
-110.096245 17.184
-130.064972 13.97
-136.111862 0.874
-277.16571 100
-289.165924 52.842
-307.177856 0.793
-406.223083 43.696
-END IONS
-
-BEGIN IONS
-NAME=Vadimezan
-DESCRIPTION=MCE bioactive compounds library
-EXACTMASS=282.089209
-FORMULA=C17H14O4
-INCHI=InChI=1S/C17H14O4/c1-9-6-7-13-15(20)12-5-3-4-11(8-14(18)19)17(12)21-16(13)10(9)2/h3-7H,8H2,1-2H3,(H,18,19)
-INCHIAUX=XGOYIMQSIKSOBS-UHFFFAOYSA-N
-SMILES=Cc1c(C)c2c(cc1)c(=O)c1cccc(CC(=O)O)c1o2
-FEATURE_ID=474
-MSLEVEL=2
-RTINSECONDS=113.94
-ADDUCT=[M+H]+
-PEPMASS=283.09648
-CHARGE=1
-Collision energy=60.0
-FRAGMENTATION_METHOD=HCD
-ISOLATION_WINDOW=1.2000000476839432
-Acquisition=Crude
-INSTRUMENT_TYPE=Orbitrap
-SOURCE_INSTRUMENT=Orbitrap ID-X
-IMS_TYPE=none
-ION_SOURCE=ESI
-IONMODE=Positive
-PI=Tomas Pluskal
-DATACOLLECTOR=Corinna Brungs
-DATASET_ID=MSVPLACEHOLDERID
-USI=mzspec:MSVPLACEHOLDERID:20220601_pluskal_mce_1D1_A2_id.mzML:474
-SCANS=474
-PRECURSOR_PURITY=0.973405752811273
-QUALITY_CHIMERIC=PASSED
-QUALITY_EXPLAINED_INTENSITY=0.99386805
-QUALITY_EXPLAINED_SIGNALS=0.9444444
-Num peaks=36
-91.053741 1.144
-105.069382 1.254
-141.069168 1.139
-152.061859 1.019
-153.069305 1.283
-155.085022 0.966
-158.033508 1.081
-165.069733 7.345
-166.077545 9.805
-167.085587 4.037
-168.093094 3.638
-178.077515 2.42
-179.085281 1.429
-181.06459 6.555
-181.101135 2.534
-183.116669 0.903
-190.078049 0.997
-191.085159 1.264
-193.101242 1.467
-194.07254 10.324
-195.080231 21.628
-196.087906 3.722
-208.088181 2.735
-209.059555 0.982
-209.096024 59.124
-209.119858 0.749
-210.104065 2.296
-221.096359 0.981
-222.067566 2.431
-223.0755 19.616
-225.091003 2.035
-236.083176 2.085
-237.090988 100
-238.098816 15.987
-239.106781 0.935
-283.096832 2.447
-END IONS