Utilizing QSAR Prediction Model to Identify Potential HER2 Inhibitors

Context: The HER2 receptor is a transmembrane protein that regulates the growth and division of cells. With other EGFR (epidermal growth factor receptor) family members, it forms homodimers or heterodimers in a ligand-dependent and -independent manner, activating signaling pathways that trigger cell proliferation, survival, and migration[1]. In some cases, if there are too many copies of the receptor, the cells will grow and divide uncontrollably and leading to more aggressive forms of cancer cells[2]. This paper anticipates the potential antagonist compounds via the Google Colab platform to evaluate their potential and suggest some HER2 inhibitors.

Objective: The research performs a ligand-based virtual screen experiment using the Artificial Neural Network-based Quantitative Structure-Activity Relationship (ANN-QSAR) to screen a large library of 50,000 molecules to identify potential HER2 inhibitors.

Conclusion: There are 2305 inactive compounds in the dataset and 1760 active compounds extracted from the CHEMBL database. These compounds have a smaller and more flexible structure when compared with other small molecule inhibitors like the (TKIs) tucatinib, lapatinib, and neratinib, which seem to have a larger and more rigid ring system [3]. These inhibitors are important to the treatment of breast cancer. Yet, Herceptin is still needed because it is more well-known as the single most important treatment for breast cancer in both ​​metastatic and neoadjuvant settings [4].