[BUG] FDR to QValue Numba Typing Error

[singjc]

Hi @RalfG,

I'm not sure if I should make the issue here or in the ms2rescore repo, but the downstream error seems to be in Mokapot.

When performing rescoring using MS2Rescore (v3.1.4) on Sage search results (Bruker immunopeptidomics data from the Thunder-DDA-PASEF paper), Mokapot (v0.10.0) seems to run into an error when computing the confidence estimates. I'm not entirely sure what the error is, but it seems to be an issue with Numba (see log below).

If you would like the sage results tsv file for debugging, let me know.

Best,

Justin

Log

2025-01-23 09:57:20 INFO     ms2rescore.feature_generators.im2deep // Running   
                             IM2Deep for PSMs from run (18/18):                 
                             `G230411_026_Slot2-29_1_9464`...                   
2025-01-23 10:10:13 INFO     ms2rescore.core // Removed 0 PSMs with rank >= 1.  
2025-01-23 10:10:40 ERROR    ms2rescore.__main__ // Failed in nopython mode     
                             pipeline (step: nopython frontend)                 
                             non-precise type array(pyobject, 1d, C)            
                             During: typing of argument at                      
                             /home/roestlab/anaconda3/envs/py310_JS/lib/python3.
                             10/site-packages/mokapot/qvalues.py (186)          
                                                                                
                             File                                               
                             "../../../../../../../../home/roestlab/anaconda3/en
                             vs/py310_JS/lib/python3.10/site-packages/mokapot/qv
                             alues.py", line 186:                               
                             def crosslink_tdc(scores, num_targets, desc=True): 
                                 <source elided>                                
                                                                                
                             @nb.njit                                           
                             ^

Full Traceback Error Log

2025-01-23 09:57:20 INFO     ms2rescore.feature_generators.im2deep // Running   
                             IM2Deep for PSMs from run (18/18):                 
                             `G230411_026_Slot2-29_1_9464`...                   
2025-01-23 10:10:13 INFO     ms2rescore.core // Removed 0 PSMs with rank >= 1.  
2025-01-23 10:10:40 ERROR    ms2rescore.__main__ // Failed in nopython mode     
                             pipeline (step: nopython frontend)                 
                             non-precise type array(pyobject, 1d, C)            
                             During: typing of argument at                      
                             /home/roestlab/anaconda3/envs/py310_JS/lib/python3.
                             10/site-packages/mokapot/qvalues.py (186)          
                                                                                
                             File                                               
                             "../../../../../../../../home/roestlab/anaconda3/en
                             vs/py310_JS/lib/python3.10/site-packages/mokapot/qv
                             alues.py", line 186:                               
                             def crosslink_tdc(scores, num_targets, desc=True): 
                                 <source elided>                                
                                                                                
                             @nb.njit                                           
                             ^
                                                                                
                             ╭─────── Traceback (most recent call last) ───────╮
                             │ /home/roestlab/anaconda3/envs/py310_JS/lib/pyth │
                             │ on3.10/site-packages/ms2rescore/__main__.py:237 │
                             │  in main                                        │
                             │                                                 │
                             │   234 │   │   │   profiled_rescore = profile(re │
                             │   235 │   │   │   profiled_rescore(configuratio │
                             │   236 │   │   else:                             │
                             │ ❱ 237 │   │   │   rescore(configuration=config) │
                             │   238 │   except Exception as e:                │
                             │   239 │   │   LOGGER.exception(e)               │
                             │   240 │   │   sys.exit(1)                       │
                             │                                                 │
                             │ /home/roestlab/anaconda3/envs/py310_JS/lib/pyth │
                             │ on3.10/site-packages/ms2rescore/core.py:168 in  │
                             │ rescore                                         │
                             │                                                 │
                             │   165 │   if all(psm_list["pep"] == 1.0):       │
                             │   166 │   │   psm_list = _fix_constant_pep(psm_ │
                             │   167 │   psm_list = _filter_by_rank(psm_list,  │
                             │ ❱ 168 │   psm_list = _calculate_confidence(psm_ │
                             │   169 │   _ = _log_id_psms_after(psm_list, id_p │
                             │       max_rank=config["max_psm_rank_output"])   │
                             │   170 │                                         │
                             │   171 │   # Write output                        │
                             │                                                 │
                             │ /home/roestlab/anaconda3/envs/py310_JS/lib/pyth │
                             │ on3.10/site-packages/ms2rescore/core.py:317 in  │
                             │ _calculate_confidence                           │
                             │                                                 │
                             │   314 │   )                                     │
                             │   315 │                                         │
                             │   316 │   # Recalculate confidence              │
                             │ ❱ 317 │   new_confidence = lin_psm_data.assign_ │
                             │   318 │                                         │
                             │   319 │   # Add new confidence estimations to P │
                             │   320 │   add_psm_confidence(psm_list, new_conf │
                             │                                                 │
                             │ /home/roestlab/anaconda3/envs/py310_JS/lib/pyth │
                             │ on3.10/site-packages/mokapot/dataset.py:607 in  │
                             │ assign_confidence                               │
                             │                                                 │
                             │   604 │   │                                     │
                             │   605 │   │   if self._group_column is None:    │
                             │   606 │   │   │   LOGGER.info("Assigning confid │
                             │ ❱ 607 │   │   │   return LinearConfidence(      │
                             │   608 │   │   │   │   self,                     │
                             │   609 │   │   │   │   scores,                   │
                             │   610 │   │   │   │   eval_fdr=eval_fdr,        │
                             │                                                 │
                             │                                                 │
                             │ /home/roestlab/anaconda3/envs/py310_JS/lib/pyth │
                             │ on3.10/site-packages/mokapot/confidence.py:375  │
                             │ in __init__                                     │
                             │                                                 │
                             │   372 │   │   self._perform_tdc(self._psm_colum │
                             │   373 │   │   LOGGER.info("\t- Found %i PSMs fr │
                             │   374 │   │                                     │
                             │ ❱ 375 │   │   self._assign_confidence(desc=desc │
                             │   376 │   │                                     │
                             │   377 │   │   self.accepted = {}                │
                             │   378 │   │   for level in self.levels:         │
                             │                                                 │
                             │ /home/roestlab/anaconda3/envs/py310_JS/lib/pyth │
                             │ on3.10/site-packages/mokapot/confidence.py:455  │
                             │ in _assign_confidence                           │
                             │                                                 │
                             │   452 │   │   │                                 │
                             │   453 │   │   │   # Estimate q-values and assig │
                             │   454 │   │   │   LOGGER.info("Assiging q-value │
                             │ ❱ 455 │   │   │   data["mokapot q-value"] = qva │
                             │   456 │   │   │                                 │
                             │   457 │   │   │   # Make output tables pretty   │
                             │   458 │   │   │   data = data.drop(self._target │
                             │                                                 │
                             │ /home/roestlab/anaconda3/envs/py310_JS/lib/pyth │
                             │ on3.10/site-packages/mokapot/qvalues.py:109 in  │
                             │ tdc                                             │
                             │                                                 │
                             │   106 │   with open('other_args.pkl', 'wb') as  │
                             │   107 │   │   pickle.dump(other_args, psm_file) │
                             │   108 │                                         │
                             │ ❱ 109 │   qvals = _fdr2qvalue(fdr, num_total, u │
                             │   110 │   qvals = np.flip(qvals)                │
                             │   111 │   qvals = qvals[np.argsort(srt_idx)]    │
                             │   112                                           │
                             │                                                 │
                             │ /home/roestlab/anaconda3/envs/py310_JS/lib/pyth │
                             │ on3.10/site-packages/numba/core/dispatcher.py:4 │
                             │ 23 in _compile_for_args                         │
                             │                                                 │
                             │    420 │   │   │   │   │      f"by the followin │
                             │    421 │   │   │   │   e.patch_message(msg)     │
                             │    422 │   │   │                                │
                             │ ❱  423 │   │   │   error_rewrite(e, 'typing')   │
                             │    424 │   │   except errors.UnsupportedError a │
                             │    425 │   │   │   # Something unsupported is p │
                             │    426 │   │   │   error_rewrite(e, 'unsupporte │
                             │                                                 │
                             │                                                 │
                             │ /home/roestlab/anaconda3/envs/py310_JS/lib/pyth │
                             │ on3.10/site-packages/numba/core/dispatcher.py:3 │
                             │ 64 in error_rewrite                             │
                             │                                                 │
                             │    361 │   │   │   if config.FULL_TRACEBACKS:   │
                             │    362 │   │   │   │   raise e                  │
                             │    363 │   │   │   else:                        │
                             │ ❱  364 │   │   │   │   raise e.with_traceback(N │
                             │    365 │   │                                    │
                             │    366 │   │   argtypes = []                    │
                             │    367 │   │   for a in args:                   │
                             ╰─────────────────────────────────────────────────╯
                             TypingError: Failed in nopython mode pipeline      
                             (step: nopython frontend)                          
                             non-precise type array(pyobject, 1d, C)            
                             During: typing of argument at                      
                             /home/roestlab/anaconda3/envs/py310_JS/lib/python3.
                             10/site-packages/mokapot/qvalues.py (186)          
                                                                                
                             File                                               
                             "../../../../../../../../home/roestlab/anaconda3/en
                             vs/py310_JS/lib/python3.10/site-packages/mokapot/qv
                             alues.py", line 186:                               
                             def crosslink_tdc(scores, num_targets, desc=True): 
                                 <source elided>                                
                                                                                
                             @nb.njit                                           
                             ^

To Reproduce:

Command

ms2rescore -p ./ms2rescore_mars/results.sage.tsv -t sage -c ./ms2rescore_mars/ms2rescore_config.json

MS2Rescore Config

{
    "$schema": "./config_schema.json",
    "ms2rescore": {
        "feature_generators": {
            "deeplc": {
                "calibration_set_size": 0.6,
                "deeplc_retrain": false
            },
            "ms2pip": {
                "model": "HCD",
                "ms2_tolerance": 0.02
            },
            "im2deep": {
                "reference_dataset": "/media/roestlab/Data1/User/JustinS/jpst002044_pxd040385_thunder_ddapasef/results/20241206_no_redeem_lda/ms2rescore_mars/reference_ccs"
            }
        },
        "rescoring_engine": {
            "mokapot": {
                "write_weights": true,
                "write_flashlfq": false
            }
        },
        "config_file": "/media/roestlab/Data1/User/JustinS/jpst002044_pxd040385_thunder_ddapasef/results/20241206_no_redeem_lda/ms2rescore_mars/ms2rescore_config.json",
        "psm_file": [
            "/media/roestlab/Data1/User/JustinS/jpst002044_pxd040385_thunder_ddapasef/results/20241206_no_redeem_lda/ms2rescore_mars//results.sage.tsv"
        ],
        "psm_file_type": "sage",
        "psm_reader_kwargs": {},
        "spectrum_path": "/media/roestlab/Data1/User/JustinS/jpst002044_pxd040385_thunder_ddapasef/results/20241206_no_redeem_lda/G230411_data/",
        "output_path": "/media/roestlab/Data1/User/JustinS/jpst002044_pxd040385_thunder_ddapasef/results/20241206_no_redeem_lda/ms2rescore_mars/",
        "log_level": "info",
        "id_decoy_pattern": "rev_",
        "psm_id_pattern": null,
        "spectrum_id_pattern": null,
        "lower_score_is_better": false,
        "modification_mapping": {
            "+15.9949": "U:Oxidation",
            "+57.0215": "U:Carbamidomethyl"
        },
        "fixed_modifications": {},
        "processes": 30,
        "rename_to_usi": false,
        "fasta_file": null,
        "write_report": true
    }
}

System

OS:

• Ubuntu 18.04.6 LTS
• 187Gb RAM
• 18 CPU cores

Python:

• Python 3.10 (Anaconda Env)
• MS2Rescore v3.1.4
• Mokapot v0.10.0

[jspaezp]

Hey there @singjc!

Thanks for reporting the error! On the next release of mokapot (which hopefully will happen in ~1 month) we will be removing the numba dependency, so this should not be a problem.

I noticed you are using py3.10 Python 3.10 (Anaconda Env), would you mind checking if py3.9 works for you?

Best,
Sebastian