<b>File 1 — Training Dataset (PubChem + ChEMBL)</b><b> for Dr Lungker</b>

<p dir="ltr"><b>Article Title:</b> <i>DrLungker: A Deep Ensemble Learning Framework for Predicting Anti-Lung Cancer Compound Activity and Validating Multitarget Potency through WaterMap, DFT, MD Simulations, and MM-GBSA Analysis</i><br><b>Published in:</b> <i>Advanced Theory and Simulations</i><br><b>Manuscript DOI:</b> <a href="https://doi.org/10.1002/adts.202501550" target="_blank">https://doi.org/10.1002/adts.202501550</a><br><b>More Information:</b> <a href="https://github.com/ShabanAhmad/DrLungker" target="_blank">GitHub Repository</a></p><h3><b>Description</b></h3><p dir="ltr">This dataset (<code>DrLungker_Dataset.csv</code>) contains the fully curated molecular data used to train the <b>DrLungker deep ensemble learning framework</b> for predicting anti-lung cancer compound activity.</p><ul><li><b>Sources:</b> PubChem and ChEMBL lung cancer bioassays</li><li><b>Processing:</b> Structure standardization, duplicate removal, descriptor generation using AlvaDesc and QikProp, and rigorous quality filtering</li><li><b>Contents:</b> 26,396 unique compounds, each encoded with 5,883 molecular descriptors</li><li><b>Usage:</b> Training the hybrid <b>ResNet–FNN–LSTM ensemble</b> using Averaging, Majority Voting, and Stacking techniques</li></ul><p dir="ltr">This dataset ensures <b>full reproducibility</b> of the DrLungker model and can be used for benchmarking, validation, and downstream computational drug-discovery applications.</p>