TACK (TArgeting Chimeras Knowledge) is a standardized knowledge base dataset and benchmark set released by the AI Laboratory for Molecular Engineering in 2026. This dataset was specifically built for the machine learning-driven PROTAC degradation activity prediction task. Related research papers include... TACK: A statistical evaluation of degradation activity on a novel TArgeting Chimeras Knowledge datasetIt aims to address the issues of scarce data, lack of rigorous evaluation, and limited coverage in existing PROTAC machine learning benchmarks. It is widely used in fields such as PROTAC degradation activity prediction, targeted protein degradation (TPD) research, AI-assisted drug discovery (AIDD), computer-aided drug design (CADD), virtual drug screening, multi-task learning, molecular property prediction, graph neural network research, and machine learning benchmark testing. This dataset contains 6,561 records, including 4,184 DC50 records, 2,377 Dmax records, and 1,563 multitask records. It encompasses 3,514 unique PROTAC molecules, 164 target proteins (POIs), 9 E3 ubiquitin ligases (E3 ligases), and 155 cell lines, exhibiting rich chemical structural features and diverse biological experimental conditions. Based on the activity criteria of DC₅₀ ≤ 100 nM and Dmax ≥ 80%, approximately 55% samples were labeled as active samples.

Dataset composition

It contains three data subsets to support different types of PROTAC degradation activity prediction tasks:

• DC50: Contains only protein degradation efficiency metrics (DC₅₀), totaling 4,184 records.
• Dmax: Includes only the maximum degradation efficiency index (Dmax) data, totaling 2,377 records.
• multitask: Contains paired DC₅₀ and Dmax data for the same PROTAC molecule under identical experimental conditions, totaling 1,563 records. Suitable for multi-task learning and binary classification studies.

Citation

@misc{ribes2026tackstatisticalevaluationdegradation,
title={TACK: A statistical evaluation of degradation activity on a novel TArgeting Chimeras Knowledge dataset},
author={Stefano Ribes and Nils Dunlop and Rocío Mercado},
year={2026},
eprint={2605.19579},
archivePrefix={arXiv},
primaryClass={q-bio.QM},
url={https://arxiv.org/abs/2605.19579},
}