APARENT2

Deep residual neural network for predicting human 3' UTR Alternative Polyadenylation (APA) and cleavage magnitude at base-pair resolution, and for deciphering the impact of genetic variants on polyadenylation.

Disclaimer

This is an UNOFFICIAL implementation of Deciphering the impact of genetic variation on human polyadenylation using APARENT2 by Johannes Linder, Samantha E. Koplik et al.

The OFFICIAL repository of APARENT2 is at johli/aparent-resnet.

The MultiMolecule team has confirmed that the provided model and checkpoints are producing the same intermediate representations as the original implementation.

The team releasing APARENT2 did not write this model card for this model so this model card has been written by the MultiMolecule team.

Model Details

APARENT2 is a residual convolutional neural network (a ResNet successor to the original APARENT) trained on a 3' UTR massively parallel reporter assay (MPRA). Given a fixed 205bp polyadenylation signal (PAS) sequence, it predicts a base-pair-resolution cleavage probability distribution as well as the overall isoform abundance. It is primarily used to score the effect of genetic variants on polyadenylation by comparing the predictions for a reference and an alternate sequence.

The network is fully convolutional: an input projection, seven groups of four pre-activation residual blocks (batch-norm → ReLU → dilated convolution, twice, with a skip connection), per-group skip convolutions that are summed, a final cleavage projection, and a position-wise locally-connected training-sub-library bias layer. There is no flatten or dense layer anywhere in the architecture.

• The core hexamer (e.g. AATAAA) is expected to start at position 70 (0-indexed) of the 205bp window.
• The model output is a 206-dimensional cleavage distribution: one score per input position plus a trailing "no cleavage in window" bucket.
• Variant effect scoring is an input-schema concern: score the reference and alternate sequences separately and compare their cleavage / isoform predictions. There is no separate ref/alt output dataclass.

Model Specification

Num Layers	Hidden Size	Num Parameters (M)	FLOPs (G)	MACs (G)	Max Num Tokens
28	32	0.19	0.08	0.04	205

Links

• Code: multimolecule.aparent2
• Weights: multimolecule/aparent2
• Paper: Deciphering the impact of genetic variation on human polyadenylation using APARENT2
• Developed by: Johannes Linder, Samantha E. Koplik, Anshul Kundaje, Georg Seelig
• Original Repository: johli/aparent-resnet

Usage

The model file depends on the multimolecule library. You can install it using pip:

pip install multimolecule

Direct Use

Polyadenylation Cleavage Prediction

You can use this model directly to predict the cleavage distribution of a 205bp polyadenylation signal sequence (core hexamer starting at position 70):

>>> import torch
>>> from multimolecule import DnaTokenizer, Aparent2Model

>>> tokenizer = DnaTokenizer.from_pretrained("multimolecule/aparent2")
>>> model = Aparent2Model.from_pretrained("multimolecule/aparent2")
>>> sequence = "A" * 70 + "AATAAA" + "A" * 129
>>> output = model(**tokenizer(sequence, return_tensors="pt"))

>>> output.logits.shape
torch.Size([1, 206])

Variant Effect Scoring

Score a reference and an alternate sequence separately, then compare:

>>> import torch
>>> ref = "A" * 70 + "AATAAA" + "A" * 129
>>> alt = "A" * 70 + "AATACA" + "A" * 129
>>> ref_prob = torch.softmax(model(**tokenizer(ref, return_tensors="pt")).logits, dim=-1)
>>> alt_prob = torch.softmax(model(**tokenizer(alt, return_tensors="pt")).logits, dim=-1)
>>> ref_iso = ref_prob[:, 77:127].sum(dim=-1)
>>> alt_iso = alt_prob[:, 77:127].sum(dim=-1)
>>> delta_logodds = torch.log(alt_iso / (1 - alt_iso)) - torch.log(ref_iso / (1 - ref_iso))

Training Details

APARENT2 was trained to predict base-pair-resolution cleavage and isoform abundance from 3' UTR MPRA measurements.

Training Data

The model was trained on the 3' UTR MPRA library used by the original APARENT, re-processed with additional improvements (exact cleavage positions for the Alien1 Random sublibrary and a 20 nt random barcode upstream of the USE in the Alien1 sublibrary). The measured variant data and processed data repository are available at the original APARENT GitHub.

Training Procedure

The model minimizes a combination of a sigmoid KL-divergence isoform loss and a KL-divergence cleavage loss, weighted equally. The released inference checkpoint corresponds to the residual-network model trained for 5 epochs on all sublibraries (excluding ClinVar wild-type sequences), with dropout disabled for inference.

Citation

@article{linder2022deciphering,
  author    = {Linder, Johannes and Koplik, Samantha E. and Kundaje, Anshul and Seelig, Georg},
  title     = {Deciphering the impact of genetic variation on human polyadenylation using APARENT2},
  journal   = {Genome Biology},
  volume    = {23},
  number    = {1},
  pages     = {232},
  year      = {2022},
  doi       = {10.1186/s13059-022-02799-4},
  publisher = {Springer Science and Business Media LLC}
}

The artifacts distributed in this repository are part of the MultiMolecule project. If you use MultiMolecule in your research, you must cite the MultiMolecule project as follows:

@software{chen_2024_12638419,
  author    = {Chen, Zhiyuan and Zhu, Sophia Y.},
  title     = {MultiMolecule},
  doi       = {10.5281/zenodo.12638419},
  publisher = {Zenodo},
  url       = {https://doi.org/10.5281/zenodo.12638419},
  year      = 2024,
  month     = may,
  day       = 4
}

Known Limitations

• The model expects a fixed 205bp input window with the core hexamer starting at position 70 (0-indexed); it does not handle variable-length sequences.
• The training-sub-library bias is fixed to the index used by the upstream variant-effect workflow.

Contact

Please use GitHub issues of MultiMolecule for any questions or comments on the model card.

Please contact the authors of the APARENT2 paper for questions or comments on the paper/model.

License

This model implementation is licensed under the GNU Affero General Public License.

For additional terms and clarifications, please refer to our License FAQ.

SPDX-License-Identifier: AGPL-3.0-or-later