Borzoi Track Activity

Score DNA target activity on selected Borzoi output tracks.

License: Borzoi uses Apache-2.0 for code and CC-BY-4.0 for model weights and may require explicit attribution when utilized. Please refer to the code license and model weights license for full terms.

This constraint is open source. Any third-party models, product names, or trademarks referenced are the property of their respective owners, and Proto is not affiliated with them.

---

Tools Used Tools Used Source Source Cite Cite

Go to Tool Page

proto-bio/proto-language/proto_language/constraint/sequence_annotation/borzoi_track_activity_constraint.py

View source

@article{linder2025borzoi,
  title={Predicting RNA-seq coverage from DNA sequence as a unifying model of gene regulation},
  author={Linder, Johannes and Srivastava, Divyanshi and Yuan, Han and Agarwal, Vikram and Kelley, David R},
  journal={Nature Genetics},
  volume={57},
  number={4},
  pages={949--961},
  year={2025},
  publisher={Nature Publishing Group},
  doi={10.1038/s41588-024-02053-6}
}

Copy citation

Score target proposals for selected Borzoi output tracks.

API Reference

ConfigBorzoiTrackActivityConfig Source

Configuration for Borzoi track-activity scoring.

organism · enum · default: "human"

Borzoi species head to use: human or mouse.

Options: human, mouse

borzoi_output_tracks · List[integer] · default: "[0]"

Borzoi output track IDs to average before scoring.

score_intervals · array

Optional target-relative intervals to score. If omitted, scores the full target segment.

direction · enum · default: "maximize"

Whether high activity or low activity is preferred.

Options: maximize, minimize

activity_threshold · number · default: "200.0"

Activity value treated as saturating success for the objective.

borzoi_ensemble_reduce_method · enum · default: "mean"

How to combine Borzoi replicate signals.

Options: mean, min, std, lcb

device · string · default: "cuda"

CUDA device for Borzoi inference.

batch_size · integer · default: "1"

Candidate sequences per Borzoi model batch.

trim_prefix_bp · integer · default: "0"

Leading target bases to ignore before Borzoi scoring.

Usage

python

from proto_language.core import Constraint
from proto_language.constraint import borzoi_track_activity_constraint, BorzoiTrackActivityConfig

constraint = Constraint(
    inputs=[segment],
    function=borzoi_track_activity_constraint,
    function_config=BorzoiTrackActivityConfig(
        # Configure parameters here
    ),
)

scores = constraint.evaluate()

Metadata

Property	Value
Key	borzoi-track-activity
Function	borzoi_track_activity_constraint
Category	sequence_annotation
Mode	discrete
Uses GPU	True
Supported Types	dna