You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
-m or --model: (required) the ChromBPNet model to use for variant scoring. For most use cases, this should be the bias-corrected model (chrombpnet_nobias.h5)
I'd be interested in a version that could make predictions on hg38 coordinates.
Thanks.
The text was updated successfully, but these errors were encountered:
Hi Peter, we should have pre-trained ChromBPNet models available soon as part of the ENCODE project. The chrombpnet_nobias.h5 file mentioned here references the output model file you get from training a ChromBPNet model on a chromatin accessibility dataset using this repo: https://github.com/kundajelab/chrombpnet. Once you have a model, you can make predictions using coordinates for any reference genome.
Where can I find
chrombpnet_nobias.h5
referenced here:variant-scorer/README.md
Line 20 in 3bd83d2
I'd be interested in a version that could make predictions on hg38 coordinates.
Thanks.
The text was updated successfully, but these errors were encountered: