search

bioinfomaticsCSU / deepsignal

Detecting methylation using signal-level features from Nanopore sequencing reads
GNU General Public License v3.0
108 stars 21 forks source link

Problems with call_mods #57

Closed cyber-ux closed 3 years ago

cyber-ux commented 3 years ago

Hi Peng, I ran the following command: "deepsignal call_mods --input_path fast5s.al.CpG.signal_features.17bases.rawsignals_360.tsv --model_path model.CpG.R9.4_1D.human_hx1.bn17.sn360.v0.1.7 + / bn_17.sn_360 .epoch_9.ckpt --result_file fast5s.al.CpG.call_mods.tsv --nproc 20 --is_gpu no "and deepsignal tells me: write_process started .. 2020-09-13 19: 30: 03.756002: I tensorflow / core / platform / cpu_feature_guard.cc: 140] Your CPU supports instructions that this TensorFlow binary was not compiled to use: AVX2 FMA 2020-09-13 19: 30: 04.019482: I tensorflow / core / platform / cpu_feature_guard.cc: 140] Your CPU supports instructions that this TensorFlow binary was not compiled to use: AVX2 FMA 2020-09-13 19: 30: 04.191046: I tensorflow / core / platform / cpu_feature_guard.cc: 140] Your CPU supports instructions that this TensorFlow binary was not compiled to use: AVX2 FMA 2020-09-13 19: 30: 04.342153: I tensorflow / core / platform / cpu_feature_guard.cc: 140] Your CPU supports instructions that this TensorFlow binary was not compiled to use: AVX2 FMA 2020-09-13 19: 30: 04.887242: I tensorflow / core / platform / cpu_feature_guard.cc: 140] Your CPU supports instructions that this TensorFlow binary was not compiled to use: AVX2 FMA 2020-09-13 19: 30: 05.001080: I tensorflow / core / platform / cpu_feature_guard.cc: 140] Your CPU supports instructions that this TensorFlow binary was not compiled to use: AVX2 FMA 2020-09-13 19: 30: 05.661715: I tensorflow / core / platform / cpu_feature_guard.cc: 140] Your CPU supports instructions that this TensorFlow binary was not compiled to use: AVX2 FMA 2020-09-13 19: 30: 05.842513: I tensorflow / core / platform / cpu_feature_guard.cc: 140] Your CPU supports instructions that this TensorFlow binary was not compiled to use: AVX2 FMA 2020-09-13 19: 30: 06.000198: I tensorflow / core / platform / cpu_feature_guard.cc: 140] Your CPU supports instructions that this TensorFlow binary was not compiled to use: AVX2 FMA 2020-09-13 19: 30: 06.043159: I tensorflow / core / platform / cpu_feature_guard.cc: 140] Your CPU supports instructions that this TensorFlow binary was not compiled to use: AVX2 FMA 2020-09-13 19: 30: 06.129178: I tensorflow / core / platform / cpu_feature_guard.cc: 140] Your CPU supports instructions that this TensorFlow binary was not compiled to use: AVX2 FMA 2020-09-13 19: 30: 06.411155: I tensorflow / core / platform / cpu_feature_guard.cc: 140] Your CPU supports instructions that this TensorFlow binary was not compiled to use: AVX2 FMA 2020-09-13 19: 30: 07.016522: I tensorflow / core / platform / cpu_feature_guard.cc: 140] Your CPU supports instructions that this TensorFlow binary was not compiled to use: AVX2 FMA 2020-09-13 19: 30: 07.065846: I tensorflow / core / platform / cpu_feature_guard.cc: 140] Your CPU supports instructions that this TensorFlow binary was not compiled to use: AVX2 FMA 2020-09-13 19: 30: 07.616329: I tensorflow / core / platform / cpu_feature_guard.cc: 140] Your CPU supports instructions that this TensorFlow binary was not compiled to use: AVX2 FMA 2020-09-13 19: 30: 08.002258: W tensorflow / core / framework / op_kernel.cc: 1318] OP_REQUIRES failed at save_restore_v2_ops.cc:184: Not found: Key BDGRU_rnn / bw / multi_rnn_cell / cell_0 / lstm_cell / bias not found in checkpoint 2020-09-13 19: 30: 08.088356: W tensorflow / core / framework / op_kernel.cc: 1318] OP_REQUIRES failed at save_restore_v2_ops.cc:184: Not found: Key BDGRU_rnn / bw / multi_rnn_cell / cell_0 / lstm_cell / bias not found in checkpoint 2020-09-13 19: 30: 08.214039: W tensorflow / core / framework / op_kernel.cc: 1318] OP_REQUIRES failed at save_restore_v2_ops.cc:184: Not found: Key BDGRU_rnn / bw / multi_rnn_cell / cell_0 / lstm_cell / bias not found in checkpoint 2020-09-13 19: 30: 08.220377: W tensorflow / core / framework / op_kernel.cc: 1318] OP_REQUIRES failed at save_restore_v2_ops.cc:184: Not found: Key BDGRU_rnn / bw / multi_rnn_cell / cell_0 / lstm_cell / bias not found in checkpoint 2020-09-13 19: 30: 08.413802: I tensorflow / core / platform / cpu_feature_guard.cc: 140] Your CPU supports instructions that this TensorFlow binary was not compiled to use: AVX2 FMA

He gave me the file.tsv but I'm not sure he executed the command correctly because it made me the errors mentioned above in a very short time.

Other exception like : " # pylint: disable=protected-access"

What can it mean?

I installed tensorflow as indicated in the instructions in a virtual environment. The current version of tensorflow is 1.8.0. I have not installed the one for the GPU because this is not that powerful and then in the "call_mods" command I set "--is_gpu no". What can it be due to then? What can I do?

PengNi commented 3 years ago

Can you check the version of deepsignal by using pip list? It seems that the version of deepsignal is lower than 0.1.7.

cyber-ux commented 3 years ago

Yes indeed it is, it tells me that the version of deepsignal is 0.1.4. What can I do ?

PengNi commented 3 years ago

You can re-install or update deepsignal to 0.1.7:

pip install deepsignal==0.1.7

Best, Peng

cyber-ux commented 3 years ago

Hi Peng, I upgraded Deepsignal but it always gives me the same thing. I now have the following versions: Tensorflow == 1.8.0 Tensorflow-estimator == 2.3.0 Tensorboard==1.8.0 Deepsignal == 0.1.7 Numpy==1.16.3 h5py==2.10.0 Scikit-learn==0.20.3 Statsmodel==0.9.0 Ont-fast5-api==3.1.5 Ont-tombo==1.5 Keras-preprocessing==1.1.2 Grpcio==1.20.0

What can I do ?

PengNi commented 3 years ago

Can you use which deepsignal to see where the bin file deepsignal is? May be the bin file is still the 1.4.0 one.

I suggest that you uninstall deepsignal and then re-install it.

Best, Peng

cyber-ux commented 3 years ago

I did not understand your question. So should I just uninstall deepsignal or also tensorflow and tombo and then reinstall them all?

PengNi commented 3 years ago

Just deepsignal.

cyber-ux commented 3 years ago

Always the version you suggested before, deepsignal 0.1.7 with the command “pip install deepsignal == 0.1.7” or just “pip install deepsignal”?

PengNi commented 3 years ago

Using

pip uninstall deepsignal
# then
pip install deepsignal==0.1.7

Also there is no space in the model path:

model.CpG.R9.4_1D.human_hx1.bn17.sn360.v0.1.7+/bn_17.sn_360.epoch_9.ckpt
cyber-ux commented 3 years ago

What do you mean by "There is also no space in the model path"

PengNi commented 3 years ago

What I mean are that make sure: 1. reinstall deepsignal; 2. the command for calling methylation is right.

Best, Peng

cyber-ux commented 3 years ago

Ok! Thank you so much Peng for helping me, I will try that and I will let you know. Thanks again. Best Regards

cyber-ux commented 3 years ago

Hi Peng I managed to make the call_mods and it was successful. I have two final questions: 1- what does the --prob_cf flag mean? 2- for the frequencies in the commands it is indicated as script "call_modification_frequency.py" but there is also another one and that is "call_modification_frequency2.py", the latter is an upload and therefore what should you use? Best Regards

PengNi commented 3 years ago

  1. --prob_cf is for ignoring ambiguous calls. If in a call abs(prob0 - prob1)<prob_cf, we abandon the call. However, --prob_cf is a experimental args, we didn't test the best value.

  2. Either call_modification_frequency.py or call_modification_frequency2.py is ok to use. call_modification_frequency2.py allows muti call_mods files as input in append mode.

Best, Peng