We've moved!
This site is now read-only. You can find our new documentation site and support forum for posting questions here.
Be sure to read our welcome blog!

CNNVariantTrain issue

Hello everyone,

I have used CNNVariantTrain tool in ( gatk to generate a 2D-model and I used the following command:
gatk CNNVariantTrain --input-tensor-dir $Read_tensor_folder --tensor-type read --model-name 2d_model

and I got the following error:
A USER ERROR has occurred: Argument tensor-type has a bad value: read-tensor. 'read-tensor' is not a valid value for TensorType. Possible values: { reference ( 1 Hot encoding of a reference sequence. ) read_tensor (Read tensor are 3D tensors spanning aligned reads, sites and channels. The maximum number of reads is a hyper-parameter typically set to 128. There are 15 channels in the read tensor. They correspond to the reference sequence data (4), read sequence data (4), insertions and deletions (2) read flags (4) and mapping quality (1).) }

can anyone help me to solve that please?


  • SakhaaSakhaa Member
    edited September 2019

    Thank you very much, I found my error it was in the syntax of read-tensor, the correct is read_tensor

    But after I corrected, I got a new error that related to generating 2D model:
    Stderr: 2019-09-17 18:28:30.043254: I tensorflow/core/platform/cpu_feature_guard.cc:141] Your CPU supports instructions that this TensorFlow binary was not compiled to use: SSE4.1 SSE4.2 AVX AVX2 AVX512F FMA 2019-09-17 18:28:30.054927: I tensorflow/core/common_runtime/process_util.cc:69] Creating new thread pool with default inter op setting: 2. Tune using inter_op_parallelism_threads for best performance. terminate called after throwing an instance of 'Xbyak::Error' what(): code is too big

  • Tiffany_at_BroadTiffany_at_Broad Cambridge, MAMember, Administrator, Broadie, Moderator admin

    Hi @Sakhaa ! I'll have to follow up with the team, as I have not run this tool myself.

    Anyone in the community seen terminate called after throwing an instance of 'Xbyak::Error' what(): code is too big ??

  • samwellsamwell Cambridge, MAMember, Broadie ✭✭
    edited September 2019

    Hi @Sakhaa, This is probably because the model architecture is too big to fit into the available memory. Try creating a smaller model by adding the argument --convLayers 32 24 for example.

  • SakhaaSakhaa Member

    Thank you @Tiffany_at_Broad and @samwell ,

    But what is the the suitable memory space to use whole layers of model?

  • bshifawbshifaw Member, Broadie, Moderator admin

    Looks like the default memory for this task in the CNN workflow is 8GB here

Sign In or Register to comment.