DeepVariant with RNASeq Model "stuck" about deepvariant

[danielecook]

@bcantarel what type of machine are you running this on? I wonder if DeepVariant is resource-constrained.

What do you get when you type nproc ? Can you try to lower the value of --num_shards to this value if it is lower?

[bcantarel]

I tried on both an e2-highmem-16 and nd2-highmem-96, so that's 16 and 96 cpus, I saw this in both instances. when running deep-variant I set num_shards to $(nproc)

[danielecook]

@bcantarel Thank you for flagging this. I was able to reproduce this issue as well. While we investigate further, you can revert back to DeepVariant 1.4. I tested v1.4 and had no issues running the RNA-seq model. Note that we have only released one RNA-seq model.

We'll continue to look into this and try to resolve it.

[kishwarshafin]

@bcantarel ,

Please use "1.5.0" for the RNA-seq model as it is trained with slim. Version 1.6.0 only works with Keras models.

[pichuan]

Hi @bcantarel ,
I hope using the right version resolved your issue. Closing this now.