Scholarlink

Background <p>The V-RAF murine sarcoma viral oncogene homolog B1 (BRAF) gene’s V600E mutation disrupts the MAP kinase/ERK-signaling pathway and is present in approximately 50% of melanoma cases. Targeting this mutation with BRAF kinase inhibitors has shown promising therapeutic results among melanoma patients, with many compounds in practice and clinical trials. Although FDA-approved BRAF inhibitors are currently available, resistance against these medications develops rapidly.</p> Methods <p>To combat this resistance crisis, a novel screening system using a machine-learning (ML) model was used to help bring additional drugs into use. Our ML model was employed to predict IC<sub>50</sub> values of FDA-approved and non-approved compounds, trained on data from experimentally characterized BRAF inhibitors. Top inhibitors were ligand-docked to the crystal structure of BRAF, and these inhibitors’ pharmacological efficacy and stability were confirmed via pharmacokinetic and ADMET analysis.</p> Results <p>The model achieved high accuracy with a Pearson correlation coefficient of 0.97, R-squared of 0.94, and RMSE of 0.25. Docking revealed a statistically significant mean difference (p-value &lt; 0.05) in binding affinity between the top- and bottom-performing inhibitors as predicted by the model. Furthermore, our top inhibitors expressed favorable pharmacokinetic profiles in ADMET properties and cytotoxicity predictions, ensuring their viability as drug candidates.</p> Discussion <p>While prior studies have used ML or docking alone to identify kinase inhibitors, they often lack integration with pharmacokinetic and safety profiling. Compared to existing approaches, our system offers two advantages: (1) it accelerates discovery by simultaneously repurposing safe, FDA-approved compounds while also uncovering novel scaffolds from large chemical libraries, and (2) it increases translational relevance by filtering candidates not only for binding affinity but also for pharmacokinetic stability and cytotoxicity, thereby reducing false positives in early-stage drug discovery. Overall, this study highlights potential drugs for future in-vitro testing and drug repurposing, underscoring the utility of ML in expanding the pool of effective BRAF inhibitors for melanoma treatment.</p> Clinical trial number <p>Not applicable.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Machine learning enabled elucidation of novel BRAF V600E inhibitors for metastatic melanoma treatment

  • 期刊论文

  • 出版模式：

    开放获取

  • 发表日期：

    2025年12月1日

查看全文
  • Aarush De,
  • Valentina L. Kouznetsova,
  • Igor F. Tsigelny

摘要

Background

The V-RAF murine sarcoma viral oncogene homolog B1 (BRAF) gene’s V600E mutation disrupts the MAP kinase/ERK-signaling pathway and is present in approximately 50% of melanoma cases. Targeting this mutation with BRAF kinase inhibitors has shown promising therapeutic results among melanoma patients, with many compounds in practice and clinical trials. Although FDA-approved BRAF inhibitors are currently available, resistance against these medications develops rapidly.

Methods

To combat this resistance crisis, a novel screening system using a machine-learning (ML) model was used to help bring additional drugs into use. Our ML model was employed to predict IC50 values of FDA-approved and non-approved compounds, trained on data from experimentally characterized BRAF inhibitors. Top inhibitors were ligand-docked to the crystal structure of BRAF, and these inhibitors’ pharmacological efficacy and stability were confirmed via pharmacokinetic and ADMET analysis.

Results

The model achieved high accuracy with a Pearson correlation coefficient of 0.97, R-squared of 0.94, and RMSE of 0.25. Docking revealed a statistically significant mean difference (p-value < 0.05) in binding affinity between the top- and bottom-performing inhibitors as predicted by the model. Furthermore, our top inhibitors expressed favorable pharmacokinetic profiles in ADMET properties and cytotoxicity predictions, ensuring their viability as drug candidates.

Discussion

While prior studies have used ML or docking alone to identify kinase inhibitors, they often lack integration with pharmacokinetic and safety profiling. Compared to existing approaches, our system offers two advantages: (1) it accelerates discovery by simultaneously repurposing safe, FDA-approved compounds while also uncovering novel scaffolds from large chemical libraries, and (2) it increases translational relevance by filtering candidates not only for binding affinity but also for pharmacokinetic stability and cytotoxicity, thereby reducing false positives in early-stage drug discovery. Overall, this study highlights potential drugs for future in-vitro testing and drug repurposing, underscoring the utility of ML in expanding the pool of effective BRAF inhibitors for melanoma treatment.

Clinical trial number

Not applicable.