There is a newer version of the record available.

Published January 14, 2026 | Version v1
Model Open

PLERIO single-protein model weights

  • 1. Faculty of Bioengineering and Bioinformatics, Lomonosov Moscow State University
  • 2. ROR icon Vavilov Institute of General Genetics

Description

This deposit contains trained model weights for PLERIO, a sequence-based framework for imputing RNA–protein interactions from eCLIP data, as described in the IJMS manuscript: "Sequence-based models for RNA-Protein interactions imputation might be insufficient for novel signal prediction in eCLIP data."

The included weights correspond to models trained on eCLIP datasets and are provided to support reproducibility and further benchmarking. These models rely solely on protein and RNA sequence features (e.g., ESM2 and UniRep embeddings) and do not incorporate structural or complex-aware information.

Note: As discussed in the paper, PLERIO’s current multiprotein inference mode is prone to false positives/negatives and should not be used for de novo prediction on unseen proteins without caution. For usage instructions, preprocessing pipelines, and evaluation scripts, please refer to the associated GitHub repositories (PLERIO repository and supporting scripts).

Files

K562_singleprot_linreg_weights.json

Files (3.9 MB)

Name Size Download all
md5:41f99d6f06167214a5f8bf37c0bd6041
3.9 MB Preview Download

Additional details

Related works

Is described by
Preprint: 10.64898/2025.12.14.694209 (DOI)

Funding

Russian Science Foundation
Сравнительный анализ больших постгеномных данных No.23-14-00136
Vavilov Institute of General Genetics
Изучение регуляции работы генов для приложений в области медицины и биотехнологий FFRW-2025-010