David Kelley

Yes, for HiC NaNs regularly occur, and we interpolate to set the values. I'm not familiar with allValidPairs format.

To make this figure, I performed the following steps. 1. Download Supp Table 6 from the DeepSEA paper from here: https://static-content.springer.com/esm/art%3A10.1038%2Fnmeth.3547/MediaObjects/41592_2015_BFnmeth3547_MOESM650_ESM.csv 2. Convert the table to a VCF file. 3....

Is it possible that you're looking at an older version of the repository? The basenji_sad command in master is different from what you listed. Check out https://github.com/calico/basenji/blob/master/tutorials/sad.ipynb The model in...

The random forest models are scikit-learn objects. They have predict functions which will take a VxT numpy array of V variants and the T Basenji tasks. For the human only...

The -b option to normalize coverage for a cutting bias model remains unfinished. If you avoid that option, you'll be on solid ground.

We had to switch the training data to requester pays because the cost of offering it was becoming far too large. I'll move the models somewhere free because they're smaller,...

OK you can now grab the models and other small files from gs://basenji_barnyard2/

I forgot to make it public. Sorry about that. Try again now

OK, I added a single tfrecord file to the public bucket gs://basenji_barnyard2/demo_tfr/train-0-0.tfr. If you need the entire dataset, just set up an account with payment and download from gs://basenji_barnyard/data/

We had to switch the training data to requester pays because the cost of offering it was becoming far too large. You'll need to setup a payment method for your...