AllenInstitute/BG-human-dilatedCNN

Dilated CNN trained on human (Homo sapiens) basal ganglia chromatin accessibility across 59 cell-type tracks. Part of the DNA sequence-modeling resources for the basal ganglia cell atlas. Published in Johansen, Fu et al., 2025.

Background

Learns cis-regulatory sequence logic underlying basal ganglia cell-type specialization from snATAC-seq accessibility profiles. Use it to score candidate enhancers, run in silico mutagenesis, or guide cell-type-specific sequence design.

Availability

The weights are public. Loading them needs RegRex (formerly enhancer-designer), an internal Allen Institute tool with a public release planned. For early access contact Kasia Kedzierska (kasia.kedzierska@alleninstitute.org).

The checkpoint names its geometry input_length and output_bins, as RegRex does since the DilatedCNN rename. The earlier format (seq_len, output_length) is kept at revision pre-input-length-rename.

Model Details

Architecture: dilated_cnn
Species: human
Number of tracks: 59
Framework: PyTorch Lightning
Checkpoint format: Lightning .ckpt

Target Tracks

AMY-SLEA-BNST_D1_GABA
AMY-SLEA-BNST_GABA
Astrocyte
BAM
BF_SKOR1_Glut
B_cells
COP
Endo
Ependymal
GPe-NDB-SI_LHX6-LHX8-GBX1_GABA
... (49 more)

Usage

import regrex as rr

model = rr.Model.from_checkpoint("bg_human_dilated_cnn")  # or "AllenInstitute/BG-human-dilatedCNN"

Downloads last month: -; Downloads are not tracked for this model. How to track

Inference Providers NEW

This model isn't deployed by any Inference Provider. 🙋 Ask for provider support