ERROR: Can only use .str accessor with string values!

[animesh]

Describe the bug
I am trying to run a 'non-specific' search of fraction of sample collected in our timTOF-pro 2 instrument against human proteome but facing this issue? I wonder if the library generated respecting non-tryptic peptides? I also see that "Lumos" is being chosen as instrument in following log, could that be issue?

To Reproduce
Following is the setup

Logs

Method Setup
CPU: 0.39%
RAM: 16.03 GB / 511.91 GB (3.13%)
0:00:00.208329 INFO: ├──optimization_manager
0:00:00.208329 INFO: │   ├──fwhm_rt: 5
0:00:00.208329 INFO: │   ├──fwhm_mobility: 0.01
0:00:00.208329 INFO: │   └──score_cutoff: 0
0:00:00.208329 INFO: ├──transfer_library
0:00:00.208329 INFO: │   ├──enabled: False
0:00:00.209329 INFO: │   ├──fragment_types:
0:00:00.209329 INFO: │   │   - b
0:00:00.209329 INFO: │   │   - y
0:00:00.209329 INFO: │   ├──max_charge: 2
0:00:00.209329 INFO: │   ├──top_k_samples: 3
0:00:00.209329 INFO: │   ├──norm_delta_max: True
0:00:00.209329 INFO: │   ├──precursor_correlation_cutoff: 0.5
0:00:00.209329 INFO: │   └──fragment_correlation_ratio: 0.75
0:00:00.209329 INFO: └──transfer_learning
0:00:00.209329 INFO:     ├──enabled: False
0:00:00.209329 INFO:     ├──batch_size: 2000
0:00:00.210329 INFO:     ├──max_lr: 0.0001
0:00:00.210329 INFO:     ├──train_fraction: 0.7
0:00:00.210329 INFO:     ├──validation_fraction: 0.2
0:00:00.210329 INFO:     ├──test_fraction: 0.1
0:00:00.210329 INFO:     ├──test_interval: 1
0:00:00.210329 INFO:     ├──lr_patience: 3
0:00:00.210329 INFO:     ├──epochs: 51
0:00:00.210329 INFO:     ├──warmup_epochs: 5
0:00:00.210329 INFO:     ├──nce: 25
0:00:00.210329 INFO:     └──instrument: Lumos
0:00:00.228420 INFO: Moved existing config file F:\promec\TIMSTOF\LARS\2025\251031_MAREN\alphaDIA\promec\library\frozen_config.yaml to F:\promec\TIMSTOF\LARS\2025\251031_MAREN\alphaDIA\promec\library\frozen_config.1.bkp.yaml
0:00:00.229421 INFO: Registering 7 custom modifications
0:00:00.307940 INFO: Searching 1 files:
0:00:00.307940 INFO:   251030_MAREN_DIALYSE_DIA_Slot1-37_1_11509.d
0:00:00.307940 INFO: Using 1 fasta files:
0:00:00.307940 INFO:   F:\promec\FastaDB\UP000005640_9606_1protein1gene.fasta
0:00:00.307940 INFO: Using library: None
0:00:00.307940 INFO: Saving output to: F:\promec\TIMSTOF\LARS\2025\251031_MAREN\alphaDIA\promec\library
0:00:00.308940 �PROGRESS: Loading spectral library�
0:00:00.308940 �PROGRESS: No library provided. Building library from fasta files.�
0:00:00.308940 INFO: Running FastaDigest
0:00:02.287389 INFO: Digesting fasta file
0:00:14.992470 INFO: Traceback (most recent call last):
  File "alphadia\cli.py", line 295, in run
    SearchPlan(output_directory, user_config, cli_params_config).run_plan()
  File "alphadia\search_plan.py", line 154, in run_plan
    self.run_step(
  File "alphadia\search_plan.py", line 193, in run_step
    step.run()
  File "alphadia\search_step.py", line 384, in run
    self.load_library()
  File "alphadia\search_step.py", line 270, in load_library
    spectral_library = fasta_digest(self.fasta_path_list)
                       ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "alphadia\libtransform\base.py", line 19, in __call__
    return self.forward(*args)
           ^^^^^^^^^^^^^^^^^^^
  File "alphadia\libtransform\fasta_digest.py", line 82, in forward
    fasta_lib.get_peptides_from_fasta_list(input)
  File "alphabase\protein\fasta.py", line 987, in get_peptides_from_fasta_list
    self.get_peptides_from_protein_dict(protein_dict)
  File "alphabase\protein\fasta.py", line 1018, in get_peptides_from_protein_dict
    self.get_peptides_from_protein_df(protein_df)
  File "alphabase\protein\fasta.py", line 998, in get_peptides_from_protein_df
    self._cleave_to_peptides(self.protein_df, protein_seq_column=digest_seq)
  File "alphabase\protein\fasta.py", line 1065, in _cleave_to_peptides
    self.refine_df()
  File "alphabase\spectral_library\base.py", line 344, in refine_df
    refine_precursor_df(self._precursor_df)
  File "alphabase\peptide\precursor.py", line 40, in refine_precursor_df
    df["nAA"] = df.sequence.str.len().astype(np.int32)
                ^^^^^^^^^^^^^^^
  File "pandas\core\generic.py", line 6299, in __getattr__
  File "pandas\core\accessor.py", line 224, in __get__
  File "pandas\core\strings\accessor.py", line 191, in __init__
  File "pandas\core\strings\accessor.py", line 245, in _validate
AttributeError: Can only use .str accessor with string values!

0:00:14.992470 �ERROR: Can only use .str accessor with string values!�
undefined

[mschwoer]

Hi,
this sounds like a bug indeed :-)

could you provide the full frozen_config.yaml and the fasta file?

best,
Magnus

[animesh]

sure @mschwoer , here is the frozen_config.yaml and UP000005640_9606_1protein1gene.zip

[mschwoer]

Thank you, I was able to reproduce your issue.

You need to set [library_prediction.missed_cleavages] to the upper [library_prediction.precursor_len] => 19 in your case.
There will be a better error message in the future: #756

Note that while analyzing, I found another issue that might affect your calculations: MannLabs/alphabase#366

let me know if it worked

best,
Magnus

[animesh]

tried to do this and hope i made the right changes @mschwoer? least this error disappears but now getting "ValueError: need at most 63 handles, got a sequence of length 66"?

Method Setup
CPU: 100.00%
RAM: 271.65 GB / 511.91 GB (53.07%)
0:00:00.177524 INFO: │   │   └──maximum_decrease_from_maximum: 0.1
0:00:00.177524 INFO: │   ├──mobility_error
0:00:00.177524 INFO: │   │   ├──targeted_update_percentile_range: 0.95
0:00:00.177524 INFO: │   │   ├──targeted_update_factor: 1.0
0:00:00.178525 INFO: │   │   ├──automatic_update_percentile_range: 0.99
0:00:00.178525 INFO: │   │   ├──automatic_update_factor: 1.1
0:00:00.178525 INFO: │   │   ├──try_narrower_values: False
0:00:00.178525 INFO: │   │   ├──maximal_decrease: 0.2
0:00:00.178525 INFO: │   │   ├──favour_narrower_optimum: False
0:00:00.178525 INFO: │   │   └──maximum_decrease_from_maximum: 0.1
0:00:00.178525 INFO: │   └──rt_error
0:00:00.178525 INFO: │       ├──targeted_update_percentile_range: 0.95
0:00:00.178525 INFO: │       ├──targeted_update_factor: 1.0
0:00:00.179524 INFO: │       ├──automatic_update_percentile_range: 0.99
0:00:00.179524 INFO: │       ├──automatic_update_factor: 1.1
0:00:00.179524 INFO: │       ├──try_narrower_values: True
0:00:00.179524 INFO: │       ├──maximal_decrease: 0.2
0:00:00.179524 INFO: │       ├──favour_narrower_optimum: True
0:00:00.179524 INFO: │       └──maximum_decrease_from_maximum: 0.1
0:00:00.179524 INFO: ├──optimization_manager
0:00:00.179524 INFO: │   ├──fwhm_rt: 5
0:00:00.179524 INFO: │   ├──fwhm_mobility: 0.01
0:00:00.180524 INFO: │   └──score_cutoff: 0
0:00:00.180524 INFO: ├──transfer_library
0:00:00.180524 INFO: │   ├──enabled: False
0:00:00.180524 INFO: │   ├──fragment_types:
0:00:00.180524 INFO: │   │   - b
0:00:00.180524 INFO: │   │   - y
0:00:00.180524 INFO: │   ├──max_charge: 2
0:00:00.180524 INFO: │   ├──top_k_samples: 3
0:00:00.180524 INFO: │   ├──norm_delta_max: True
0:00:00.181523 INFO: │   ├──precursor_correlation_cutoff: 0.5
0:00:00.181523 INFO: │   └──fragment_correlation_ratio: 0.75
0:00:00.181523 INFO: └──transfer_learning
0:00:00.181523 INFO:     ├──enabled: False
0:00:00.181523 INFO:     ├──batch_size: 2000
0:00:00.181523 INFO:     ├──max_lr: 0.0001
0:00:00.181523 INFO:     ├──train_fraction: 0.7
0:00:00.181523 INFO:     ├──validation_fraction: 0.2
0:00:00.181523 INFO:     ├──test_fraction: 0.1
0:00:00.182525 INFO:     ├──test_interval: 1
0:00:00.182525 INFO:     ├──lr_patience: 3
0:00:00.182525 INFO:     ├──epochs: 51
0:00:00.182525 INFO:     ├──warmup_epochs: 5
0:00:00.182525 INFO:     ├──nce: 25
0:00:00.182525 INFO:     └──instrument: Lumos
0:00:00.200517 INFO: Moved existing config file F:\promec\TIMSTOF\LARS\2025\251031_MAREN\alphaDIA\promec\library\frozen_config.yaml to F:\promec\TIMSTOF\LARS\2025\251031_MAREN\alphaDIA\promec\library\frozen_config.2.bkp.yaml
0:00:00.200517 INFO: Registering 7 custom modifications
0:00:00.280586 INFO: Searching 1 files:
0:00:00.280586 INFO:   251030_MAREN_DIALYSE_DIA_Slot1-37_1_11509.d
0:00:00.280586 INFO: Using 1 fasta files:
0:00:00.281587 INFO:   F:\promec\FastaDB\UP000005640_9606_1protein1gene.fasta
0:00:00.281587 INFO: Using library: None
0:00:00.281587 INFO: Saving output to: F:\promec\TIMSTOF\LARS\2025\251031_MAREN\alphaDIA\promec\library
0:00:00.281587 �PROGRESS: Loading spectral library�
0:00:00.281587 �PROGRESS: No library provided. Building library from fasta files.�
0:00:00.281587 INFO: Running FastaDigest
0:00:02.293577 INFO: Digesting fasta file
0:04:05.201411 INFO: Adding modifications
1:27:32.135095 INFO: Removing non-canonical amino acids
1:34:44.878408 INFO: Fasta library contains 185,905,081 precursors
1:34:44.932939 �PROGRESS: Predicting library properties.�
1:34:44.932939 INFO: Running PeptDeepPrediction
1:34:44.933939 INFO: Device set to cpu
1:34:52.476180 INFO: Loading PeptDeep models of type generic
1:34:54.325094 INFO: Predicting RT, MS2 and mobility
1:34:54.326095 INFO: Using multiprocessing with 64 processes ...
1:34:56.898140 INFO: Predicting rt,ms2,mobility ...
Exception in thread Thread-2 (_handle_workers):
Traceback (most recent call last):
  File "threading.py", line 1045, in _bootstrap_inner
  File "threading.py", line 982, in run
  File "multiprocessing\pool.py", line 522, in _handle_workers
  File "multiprocessing\pool.py", line 502, in _wait_for_updates
  File "multiprocessing\connection.py", line 896, in wait
  File "multiprocessing\connection.py", line 828, in _exhaustive_wait
ValueError: need at most 63 handles, got a sequence of length 66

[GeorgWa]

I will have a look into the issue. Please be advised that timsTOF data is not yet supported by the much faster Rust backend. Searching large libraries with non-specific digests will take very long and not give good results with alphaDIA. I would recommend starting with Sage or MSFragger and generating a spectral library first.

Best,
Georg

[mschwoer]

The ValueError: need at most 63 handles, got a sequence of length 66 comes from a Windows limitation when it comes to python multiprocessing, cf. https://stackoverflow.com/questions/65252807/multiprocessing-pool-pool-on-windows-cpu-limit-of-63)

We'll add a warning message when too many threads are used in Windows.

Try reducing the number of threads to 60 to proceed (but @GeorgWa's comment if ofc still valid, I was also thinking that this might be library a bit too big ;-) )

best,
Magnus

[animesh]

Thanks guys for the blazing support! I am actually not sure how to generate library with Sage/MSFragger, will those MaxQuant evidence/msms.tsv also work as libraries?