AttributeError: 'BayesNet' object has no attribute 'train_stats'

[Marwansha]

i am trying to run MOFA on my mudata object and i run to this issue

MuData object with n_obs × n_vars = 2 × 698441
  var:	'mean', 'total', 'tech', 'bio', 'FDR', 'p.value', 'hvg', 'highly_variable', 'std'
  uns:	'psbulk_stats'
  38 modalities
    MONO.CD14:	2 x 26641
      var:	'mean', 'total', 'tech', 'bio', 'FDR', 'p.value', 'hvg', 'highly_variable', 'std'
      uns:	'log1p'
      layers:	'psbulk_props'
      

m=mu.tl.mofa(mdata,
           use_obs='union',
           convergence_mode='medium',
           n_factors=2,


           use_var='highly_variable'
           )
           

        #########################################################
        ###           __  __  ____  ______                    ### 
        ###          |  \/  |/ __ \|  ____/\    _             ### 
        ###          | \  / | |  | | |__ /  \ _| |_           ### 
        ###          | |\/| | |  | |  __/ /\ \_   _|          ###
        ###          | |  | | |__| | | / ____ \|_|            ###
        ###          |_|  |_|\____/|_|/_/    \_\              ###
        ###                                                   ### 
        ######################################################### 
       
 
        
Loaded view='MONO.CD14' group='group1' with N=2 samples and D=14177 features...
Loaded view='Plasma cells' group='group1' with N=2 samples and D=11134 features...
Loaded view='MONO.CD16' group='group1' with N=2 samples and D=13649 features...
Loaded view='pDC' group='group1' with N=2 samples and D=12461 features...
Loaded view='B.Switched.memory' group='group1' with N=2 samples and D=14037 features...
Loaded view='B.Non.Switched.memory' group='group1' with N=2 samples and D=13939 features...
Loaded view='Progenitors' group='group1' with N=2 samples and D=13104 features...
Loaded view='cDC' group='group1' with N=2 samples and D=13520 features...
Loaded view='B.Naive' group='group1' with N=2 samples and D=13797 features...
Loaded view='B.Exhausted memory' group='group1' with N=2 samples and D=13326 features...
Loaded view='T.CD8.N' group='group1' with N=2 samples and D=13902 features...
Loaded view='NK.brt' group='group1' with N=2 samples and D=13868 features...
Loaded view='T.CD4.N' group='group1' with N=2 samples and D=13999 features...
Loaded view='T.CD4.CM.Tfh' group='group1' with N=2 samples and D=13948 features...
Loaded view='T.CD4.EM.TH22' group='group1' with N=2 samples and D=13673 features...
Loaded view='NK.dim' group='group1' with N=2 samples and D=13956 features...
Loaded view='T.CD4.EM.TH2' group='group1' with N=2 samples and D=13707 features...
Loaded view='T.CD4.EM.TH17' group='group1' with N=2 samples and D=13813 features...
Loaded view='T.CD4.EM.TH1/17' group='group1' with N=2 samples and D=13901 features...
Loaded view='T.CD8.CM' group='group1' with N=2 samples and D=13863 features...
Loaded view='T.Reg.Memory' group='group1' with N=2 samples and D=13616 features...
Loaded view='T.CD8.TEM' group='group1' with N=2 samples and D=13929 features...
Loaded view='CD5 B cells' group='group1' with N=2 samples and D=11382 features...
Loaded view='T.CD4.E.MIR155' group='group1' with N=2 samples and D=11628 features...
Loaded view='T.Reg.KLRB1' group='group1' with N=2 samples and D=12465 features...
Loaded view='MAIT' group='group1' with N=2 samples and D=13859 features...
Loaded view='T.CD8.CM.CCR4' group='group1' with N=2 samples and D=13741 features...
Loaded view='T.gd' group='group1' with N=2 samples and D=13681 features...
Loaded view='T.CD8.EMRA' group='group1' with N=2 samples and D=13724 features...
Loaded view='T.Reg.N' group='group1' with N=2 samples and D=13637 features...
Loaded view='MONO.Intermediate' group='group1' with N=2 samples and D=8326 features...
Loaded view='T.DN' group='group1' with N=2 samples and D=9545 features...
Loaded view='T.MEM.KLRC2' group='group1' with N=2 samples and D=10508 features...
Loaded view='T.CD4.CTL' group='group1' with N=2 samples and D=13573 features...
Loaded view='T.CD8.NKT' group='group1' with N=2 samples and D=12777 features...
Loaded view='NK.M.LIKE' group='group1' with N=2 samples and D=13106 features...
Loaded view='unassigned' group='group1' with N=2 samples and D=12665 features...
Loaded view='ILC' group='group1' with N=2 samples and D=8174 features...


Warning: 1 features(s) in view 4 have zero variance, consider removing them before training the model...

Model options:
- Automatic Relevance Determination prior on the factors: True
- Automatic Relevance Determination prior on the weights: True
- Spike-and-slab prior on the factors: False
- Spike-and-slab prior on the weights: True
Likelihoods:
- View 0 (MONO.CD14): gaussian
- View 1 (Plasma cells): gaussian
- View 2 (MONO.CD16): gaussian
- View 3 (pDC): gaussian
- View 4 (B.Switched.memory): gaussian
- View 5 (B.Non.Switched.memory): gaussian
- View 6 (Progenitors): gaussian
- View 7 (cDC): gaussian
- View 8 (B.Naive): gaussian
- View 9 (B.Exhausted memory): gaussian
- View 10 (T.CD8.N): gaussian
- View 11 (NK.brt): gaussian
- View 12 (T.CD4.N): gaussian
- View 13 (T.CD4.CM.Tfh): gaussian
- View 14 (T.CD4.EM.TH22): gaussian
- View 15 (NK.dim): gaussian
- View 16 (T.CD4.EM.TH2): gaussian
- View 17 (T.CD4.EM.TH17): gaussian
- View 18 (T.CD4.EM.TH1/17): gaussian
- View 19 (T.CD8.CM): gaussian
- View 20 (T.Reg.Memory): gaussian
- View 21 (T.CD8.TEM): gaussian
- View 22 (CD5 B cells): gaussian
- View 23 (T.CD4.E.MIR155): gaussian
- View 24 (T.Reg.KLRB1): gaussian
- View 25 (MAIT): gaussian
- View 26 (T.CD8.CM.CCR4): gaussian
- View 27 (T.gd): gaussian
- View 28 (T.CD8.EMRA): gaussian
- View 29 (T.Reg.N): gaussian
- View 30 (MONO.Intermediate): gaussian
- View 31 (T.DN): gaussian
- View 32 (T.MEM.KLRC2): gaussian
- View 33 (T.CD4.CTL): gaussian
- View 34 (T.CD8.NKT): gaussian
- View 35 (NK.M.LIKE): gaussian
- View 36 (unassigned): gaussian
- View 37 (ILC): gaussian



Warning: some group(s) have less than 15 samples, MOFA won't be able to learn meaningful factors for these group(s)...



######################################
## Training the model with seed 1 ##
######################################



Converged!

Attempting to save the model at the current iteration...
Saving model in /tmp/mofa_20240214-141144_interrupted.hdf5...
---------------------------------------------------------------------------
TypeError                                 Traceback (most recent call last)
File ~/venvs/scvi/lib/python3.11/site-packages/mofapy2/run/entry_point.py:57, in keyboardinterrupt_saver.<locals>.saver(self, *args, **kwargs)
     56 try:
---> 57     func(self, *args, **kwargs)
     58 # Internal methods will raise TypeError when interrupted

File ~/venvs/scvi/lib/python3.11/site-packages/mofapy2/run/entry_point.py:1434, in entry_point.run(self)
   1433 # Train the model
-> 1434 train_model(self.model)

File ~/venvs/scvi/lib/python3.11/site-packages/mofapy2/build_model/train_model.py:27, in train_model(model)
     25 print("\n")
---> 27 model.iterate()
     29 print("\n")

File ~/venvs/scvi/lib/python3.11/site-packages/mofapy2/core/BayesNet.py:387, in BayesNet.iterate(self)
    381 finally:
    382     # Finish by collecting the training statistics
    383     self.train_stats = {
    384         "time": iter_time,
    385         "number_factors": number_factors,
    386         "elbo": elbo["total"].values,
--> 387         "elbo_terms": elbo.drop("total", 1),
    388     }
    389     if "Sigma" in self.nodes.keys():

TypeError: DataFrame.drop() takes from 1 to 2 positional arguments but 3 were given

During handling of the above exception, another exception occurred:

AttributeError                            Traceback (most recent call last)
Cell In[56], line 1
----> 1 m=mu.tl.mofa(mdata,
      2            use_obs='union',
      3            convergence_mode='medium',
      4            n_factors=2,
      5 
      6 
      7            use_var='highly_variable'
      8            )

File ~/venvs/scvi/lib/python3.11/site-packages/muon/_core/tools.py:588, in mofa(data, groups_label, use_raw, use_layer, use_var, use_obs, likelihoods, n_factors, scale_views, scale_groups, center_groups, ard_weights, ard_factors, spikeslab_weights, spikeslab_factors, n_iterations, convergence_mode, use_float32, gpu_mode, gpu_device, svi_mode, svi_batch_size, svi_learning_rate, svi_forgetting_rate, svi_start_stochastic, smooth_covariate, smooth_warping, smooth_kwargs, save_parameters, save_data, save_metadata, seed, outfile, expectations, save_interrupted, verbose, quiet, copy)
    586 ent.build()
    587 logging.info(f"[{datetime.now().strftime('%Y-%m-%d %H:%M:%S')}] Running the model...")
--> 588 ent.run()
    590 if (
    591     smooth_kwargs is not None
    592     and "new_values" in smooth_kwargs
    593     and smooth_kwargs["new_values"]
    594     and smooth_covariate
    595 ):
    596     logging.info(f"[{datetime.now().strftime('%Y-%m-%d %H:%M:%S')}] Interpolating factors...")

File ~/venvs/scvi/lib/python3.11/site-packages/mofapy2/run/entry_point.py:70, in keyboardinterrupt_saver.<locals>.saver(self, *args, **kwargs)
     65     else:
     66         tmp_file = os.path.join(
     67             "/tmp",
     68             "mofa_{}_interrupted.hdf5".format(strftime("%Y%m%d-%H%M%S")),
     69         )
---> 70     self.save(outfile=tmp_file)
     71     print(
     72         "Saved partially trained model in {}. Exiting now.".format(tmp_file)
     73     )
     74 else:

File ~/venvs/scvi/lib/python3.11/site-packages/mofapy2/run/entry_point.py:1778, in entry_point.save(self, outfile, save_data, save_parameters, expectations)
   1775     tmp.saveSmoothOptions(self.smooth_opts)
   1777 # Save training statistics
-> 1778 tmp.saveTrainingStats()
   1780 # Save variance explained values
   1781 tmp.saveVarianceExplained()

File ~/venvs/scvi/lib/python3.11/site-packages/mofapy2/build_model/save_model.py:754, in saveModel.saveTrainingStats(self)
    751 """Method to save the training statistics"""
    753 # Get training statistics
--> 754 stats = self.model.getTrainingStats()
    756 # Create HDF5 group
    757 stats_grp = self.hdf5.create_group("training_stats")

File ~/venvs/scvi/lib/python3.11/site-packages/mofapy2/core/BayesNet.py:503, in BayesNet.getTrainingStats(self)
    501 def getTrainingStats(self):
    502     """Method to return training statistics"""
--> 503     return self.train_stats

AttributeError: 'BayesNet' object has no attribute 'train_stats'

herre is the session info

anndata 0.10.3
decoupler 1.5.0
liana 1.0.4
matplotlib 3.8.1
mofapy2 0.7.0
mudata 0.2.3
muon 0.1.5
numpy 1.26.1
pandas 2.1.1
plotnine 0.12.4
scanpy 1.9.6
seaborn 0.12.2
session_info 1.0.0

Click to view modules imported as dependencies

IPython 8.17.2
jupyter_client 8.6.0
jupyter_core 5.5.0

Python 3.11.4 (main, Jul 5 2023, 13:45:01) [GCC 11.2.0]
Linux-4.18.0-477.43.1.el8_8.x86_64-x86_64-with-glibc2.28

Session information updated at 2024-02-14 14:19

[dbdimitrov]

Hi @Marwansha, it seems like this is a bug with mofapy2 0.7.0, if you update to 0.7.1 it should be resolved. Works for me at least :)

[Connorr0]

I am using 0.7.1 and ran into the same problem