README.md

metadata

license: cc-by-4.0
library_name: bpnet
tags:
  - bpnet
  - dna
  - genomics
  - transcription-factor-binding
  - encode
  - ChIP-seq
  - hg38
  - qc-passed
  - JUNB

ENCODE BPNet -- JUNB ChIP-seq in A549 (ENCSR431LRW)

Trained BPNet model (ChIP-seq) from the ENCODE project.

• Experiment: ENCSR431LRW
• Model annotation: ENCSR899KPV
• Assembly: hg38 · Target: JUNB · Biosample: A549

QC

• Status: passed
• Notes: Found direct motif (counts, profile);

Files

5-fold cross-validation. Each fold_*/ holds the trained model in two forms:

• model.h5 — Keras weights (needs the bpnet custom layer to load)
• saved_model/ — TensorFlow SavedModel (portable; loads with no extra deps)

Load

from huggingface_hub import snapshot_download
import tensorflow as tf
d = snapshot_download("kundajelab/encode-bpnet-JUNB-ChIP-seq-A549-ENCSR431LRW-ENCSR899KPV")
model = tf.saved_model.load(f"{d}/fold_0/saved_model")  # portable
# Keras .h5 (needs the bpnet package):
# from bpnet.model.custommodel import CustomModel
# m = tf.keras.models.load_model(f"{d}/fold_0/model.h5",
#         custom_objects={'CustomModel': CustomModel})

Inference inputs

The serving_default signature takes three inputs (not sequence alone):

• sequence — one-hot DNA, shape (N, 2114, 4)
• profile_bias_input_0 — control (bias) profile track, shape (N, 1000, 2)
• counts_bias_input_0 — control log-count(s), shape (N, 2)

The bias inputs are the experiment's matched control signal (the model file's derived_from control bigWigs on the ENCODE portal). Outputs: profile_predictions (N, 1000, 2) and logcounts_predictions (N, 1). Reverse-complement averaging is the production default.

License & citation

Released under CC-BY-4.0, matching the ENCODE data-use policy. Please cite the ENCODE Project Consortium and the model software: BPNet (Avsec et al., Nat Genet 2021).