Prepec modification error: Multiple modifications per site not supported in Peptide Record format.

[NicolasProvencher]

Hi, I am trying to use ms2Rescore to rescore the output i get from a searchGUi / peptideshaker analysis here is the log of ms2rescore in debug mode

Running DeepLC for PSMs from run (1/1): `586APMS_FLAG_alt1`...
Calibrating DeepLC...
Using 5135 PSMs for calibration
Multiple modifications per site not supported in Peptide Record format.
Traceback (most recent call last):
  File "ms2rescore\gui\function2ctk.py", line 301, in run
    self.fn(*self.fn_args, **self.fn_kwargs)
  File "ms2rescore\gui\app.py", line 637, in function
    rescore(configuration=config)
  File "ms2rescore\core.py", line 76, in rescore
    fgen.add_features(psm_list)
  File "ms2rescore\feature_generators\deeplc.py", line 163, in add_features
    seq_df=self._psm_list_to_deeplc_peprec(psm_list_calibration)
           ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "ms2rescore\feature_generators\deeplc.py", line 210, in _psm_list_to_deeplc_peprec
    peprec = peptide_record.to_dataframe(psm_list)
             ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "psm_utils\io\peptide_record.py", line 505, in to_dataframe
    return pd.DataFrame([PeptideRecordWriter._psm_to_entry(psm) for psm in psm_list])
                        ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "psm_utils\io\peptide_record.py", line 505, in <listcomp>
    return pd.DataFrame([PeptideRecordWriter._psm_to_entry(psm) for psm in psm_list])
                         ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "psm_utils\io\peptide_record.py", line 285, in _psm_to_entry
    sequence, modifications, charge = proforma_to_peprec(psm.peptidoform)
                                      ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "psm_utils\io\peptide_record.py", line 443, in proforma_to_peprec
    ms2pip_mods.append(_mod_to_ms2pip(mod, i + 1))
                       ^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "psm_utils\io\peptide_record.py", line 433, in _mod_to_ms2pip
    raise InvalidPeprecModificationError(
psm_utils.io.peptide_record.InvalidPeprecModificationError: Multiple modifications per site not supported in Peptide Record format.

On searchGUI i used the thermofisherrawparser to parse my raw file, X!Tandem and MS-GF+ as search algorithms and used peptideshaker In peptideshaker, i used the export->peptideshaker project as -> mzldentML to produce the mzid im using in ms2rescore (including protein sequences)

modification used are Carbomidomethyl of C as fixed, and Oxidation , Acetylation for variables

percolator was used but i dont think it got to its step

I parsed the peptides in the mzid file to make sure that for the same peptide, all the modification had a different location

here is a link to a google drive folder containing the mzid and mzml and the fasta i used https://drive.google.com/drive/folders/1DUUn7fyeJR2rgIze2dQQ3r4ikcyfwiLI?usp=sharing

here is the line in the config.json file generated by ms2rescore

`{
    "$schema": "./config_schema.json",
    "ms2rescore": {
        "feature_generators": {
            "basic": {},
            "ms2pip": {
                "model": "HCD2021",
                "ms2_tolerance": 0.02
            },
            "deeplc": {
                "deeplc_retrain": false,
                "n_epochs": 20,
                "calibration_set_size": 0.15
            }
        },
        "rescoring_engine": {
            "percolator": {
                "write_weights": true,
                "write_txt": true,
                "write_flashlfq": false,
                "protein_kwargs": {}
            }
        },
        "config_file": null,
        "psm_file": [
            "C:/Users/pron2107/Desktop/mspipeline/test1/report/test1.mzid"
        ],
        "psm_file_type": "mzid",
        "psm_reader_kwargs": {},
        "spectrum_path": "C:/Users/pron2107/Desktop/mspipeline/test1/586APMS_FLAG_alt1.mzML",
        "output_path": "C:/Users/pron2107/Desktop/mspipeline/test1/ms2rescore/test1",
        "log_level": "info",
        "id_decoy_pattern": "_REVERSED",
        "psm_id_pattern": ".*scan=(\\d+)$",
        "spectrum_id_pattern": ".*scan=(\\d+)$",
        "lower_score_is_better": false,
        "modification_mapping": {
            "Oxidation": "U:Oxidation",
            "Acetyl": "U:Acetylation"
        },
        "fixed_modifications": {
            "U:Carbamidomethyl": [
                "C"
            ]
        },
        "processes": 12,
        "rename_to_usi": false,
        "fasta_file": "C:/Users/pron2107/Desktop/mspipeline/test1/human-openprot-2_0-refprots+altprots+isoforms-min_2_pep-uniprot2022_06_011_concatenated_target_decoy.fasta",
        "write_report": true
    }
}`

after all that i am wondering if I am doing anything wrong, with my settings

Thanks in advance

Nicolas

[RalfG]

Hi Nicolas,

Apologies in the delay in getting back to you. Your issue seems to be a combination of 2 problems:

• We have yet to update the DeepLC integration to send a PSMList object straight to DeepLC instead of first converting it to the old peprec format (opened an issue for that #113)
• The configuration for fixed_modifications should be empty, as they are already imported in the mzIdentML file. That is now more clearly described in the documentation. Just updating the config as follows should fix the issue:

       "fixed_modifications": {},
  

Let me know if it works or not.

Best, Ralf

[NicolasProvencher]

It works but i get another error

Extracting CNN features
Time to calculate all features: 2.843510389328003 seconds
got feature extraction results
Creating converter from 3 to 5
'NoneType' object has no attribute 'write'
Traceback (most recent call last):
  File "ms2rescore\gui\function2ctk.py", line 301, in run
    self.fn(*self.fn_args, **self.fn_kwargs)
  File "ms2rescore\gui\app.py", line 637, in function
    rescore(configuration=config)
  File "ms2rescore\core.py", line 76, in rescore
    fgen.add_features(psm_list)
  File "ms2rescore\feature_generators\deeplc.py", line 162, in add_features
    self.deeplc_predictor.calibrate_preds(
  File "deeplc\deeplc.py", line 1005, in calibrate_preds
    calibrate_output = self.calibrate_preds_func_pygam(
                       ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "deeplc\deeplc.py", line 712, in calibrate_preds_func_pygam
    predicted_tr = self.make_preds(
                   ^^^^^^^^^^^^^^^^
  File "deeplc\deeplc.py", line 667, in make_preds
    ret_preds = self.make_preds_core(X=X,
               ^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "deeplc\deeplc.py", line 578, in make_preds_core
    ret_preds = mod.predict(
                ^^^^^^^^^^^^
  File "keras\utils\traceback_utils.py", line 70, in error_handler
  File "keras\utils\traceback_utils.py", line 67, in error_handler
  File "keras\engine\training.py", line 2403, in predict
  File "keras\callbacks.py", line 519, in on_predict_batch_end
  File "keras\callbacks.py", line 322, in _call_batch_hook
  File "keras\callbacks.py", line 345, in _call_batch_end_hook
  File "keras\callbacks.py", line 393, in _call_batch_hook_helper
  File "keras\callbacks.py", line 1101, in on_predict_batch_end
  File "keras\callbacks.py", line 1170, in _batch_update_progbar
  File "keras\utils\generic_utils.py", line 296, in update
  File "keras\utils\io_utils.py", line 79, in print_msg
AttributeError: 'NoneType' object has no attribute 'write'

This time i dont have any clue where to start looking

[RalfG]

Picking up on this issue again. Apologies for the long delay.

@RobbinBouwmeester, any ideas?

@NicolasProvencher, could you check if updating DeepLC fixes the issue (pip install -U deeplc). Thanks!

[NicolasProvencher]

Hi, ralph Hope you are doing well,

First, since I was testing your windows installation to make a protocol for a gui user, things like pip are not used.

Now, I just tried reinstalling the .exe file for version 3.0.3 and the program wont open after the installation for me to retest everything and give you more info heres the error i encounter right now'

if i scroll at the end i see

Click to expand

Traceback (most recent call last): File "ms2rescore\gui\__main__.py", line 7, in <module> from ms2rescore.gui.app import app File "PyInstaller\loader\pyimod02_importers.py", line 419, in exec_module File "ms2rescore\__init__.py", line 16, in <module> from ms2rescore.core import rescore # noqa: F401 E402 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ File "PyInstaller\loader\pyimod02_importers.py", line 419, in exec_module File "ms2rescore\core.py", line 9, in <module> from ms2rescore.feature_generators import FEATURE_GENERATORS File "PyInstaller\loader\pyimod02_importers.py", line 419, in exec_module File "ms2rescore\feature_generators\__init__.py", line 9, in <module> from ms2rescore.feature_generators.ms2pip import MS2PIPFeatureGenerator File "PyInstaller\loader\pyimod02_importers.py", line 419, in exec_module File "ms2rescore\feature_generators\ms2pip.py", line 34, in <module> from ms2pip import correlate File "PyInstaller\loader\pyimod02_importers.py", line 419, in exec_module File "ms2pip\__init__.py", line 13, in <module> from ms2pip.core import ( # noqa: F401 E402 File "PyInstaller\loader\pyimod02_importers.py", line 419, in exec_module File "ms2pip\core.py", line 25, in <module> from ms2pip._utils.xgb_models import get_predictions_xgb, validate_requested_xgb_model File "PyInstaller\loader\pyimod02_importers.py", line 419, in exec_module File "ms2pip\_utils\xgb_models.py", line 10, in <module> import xgboost as xgb File "PyInstaller\loader\pyimod02_importers.py", line 419, in exec_module File "xgboost\__init__.py", line 7, in <module> from . import collective, dask, rabit File "PyInstaller\loader\pyimod02_importers.py", line 419, in exec_module File "xgboost\dask.py", line 77, in <module> from .sklearn import ( File "PyInstaller\loader\pyimod02_importers.py", line 419, in exec_module File "xgboost\sklearn.py", line 22, in <module> from scipy.special import softmax File "PyInstaller\loader\pyimod02_importers.py", line 419, in exec_module File "scipy\special\__init__.py", line 777, in <module> File "scipy\\special\\_ufuncs.pyx", line 1, in init scipy.special._ufuncs ModuleNotFoundError: No module named 'scipy.special._cdflib'

If youd like me to open a different issue for this particuliar issue, just ask

If testing and fixing the windows gui version isnt a priority right now I am ok to wait until it is

Thanks a lot

[RalfG]

Hi @NicolasProvencher,

Thanks for reporting! I opened a new issue here: #145

Best, Ralf

[NicolasProvencher]

Hi, I am following up on error given while trying to run deeplc using the windows interface. we redid a run to see if the updates that happened since the original posting fixed the problem. It seems like it did not, I included the traceback we get when it crashes.

Collapsible Section Title

Calibrating DeepLC...
Using 6036 PSMs for calibration
Start to calibrate predictions ...
Ready to find the best model out of: ['C:\\Users\\jacj2401\\AppData\\Local\\Programs\\MS2Rescore\\_internal\\deeplc\\mods/full_hc_PXD005573_mcp_1fd8363d9af9dcad3be7553c39396960.hdf5']
Trying out the following model: C:\Users\jacj2401\AppData\Local\Programs\MS2Rescore\_internal\deeplc\mods/full_hc_PXD005573_mcp_1fd8363d9af9dcad3be7553c39396960.hdf5
Extracting features for the CNN model ...
prepare feature extraction
start feature extraction
wait for feature extraction
get feature extraction results
got feature extraction results
Creating converter from 3 to 5
'NoneType' object has no attribute 'write'
Traceback (most recent call last):
  File "ms2rescore\gui\function2ctk.py", line 301, in run
    self.fn(*self.fn_args, **self.fn_kwargs)
  File "ms2rescore\gui\app.py", line 638, in function
    rescore(configuration=config)
  File "ms2rescore\core.py", line 80, in rescore
    fgen.add_features(psm_list)
  File "ms2rescore\feature_generators\deeplc.py", line 162, in add_features
    self.deeplc_predictor.calibrate_preds(
  File "deeplc\deeplc.py", line 1173, in calibrate_preds
    calibrate_output = self.calibrate_preds_func_pygam(
                       ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "deeplc\deeplc.py", line 830, in calibrate_preds_func_pygam
    predicted_tr = self.make_preds(psm_list, calibrate=False, mod_name=mod_name)
                   ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "deeplc\deeplc.py", line 750, in make_preds
    ret_preds = self.make_preds_core(
                ^^^^^^^^^^^^^^^^^^^^^
  File "deeplc\deeplc.py", line 621, in make_preds_core
    ret_preds = mod.predict(
                ^^^^^^^^^^^^
  File "keras\utils\traceback_utils.py", line 70, in error_handler
  File "keras\utils\traceback_utils.py", line 67, in error_handler
  File "keras\engine\training.py", line 2403, in predict
  File "keras\callbacks.py", line 519, in on_predict_batch_end
  File "keras\callbacks.py", line 322, in _call_batch_hook
  File "keras\callbacks.py", line 345, in _call_batch_end_hook
  File "keras\callbacks.py", line 393, in _call_batch_hook_helper
  File "keras\callbacks.py", line 1101, in on_predict_batch_end
  File "keras\callbacks.py", line 1170, in _batch_update_progbar
  File "keras\utils\generic_utils.py", line 296, in update
  File "keras\utils\io_utils.py", line 79, in print_msg
AttributeError: 'NoneType' object has no attribute **'write'**

If the windows version of ms2rescore isnt a priority right now feel free to bump down the importance of this issues since we are not actively trying to use it right now.

Best Nicolas