Influenza virus is known for its intermittent outbreaks affecting billions of people worldwide. Several neuraminidase inhibitors have been used in practice to overcome this situation. However, advent of new resistant mutants has limited its clinical utilization. In the recent years drug repurposing technique has attained the limelight as it is cost effective and reduces the time consumed for drug discovery. Here, we present multi-dimensional repurposing strategy that integrates the results of ligand-, energy-, receptor cavity, and shape-based pharmacophore algorithm to effectively identify novel drug candidate for influenza. The pharmacophore hypotheses were generated by utilizing the PHASE module of Schrödinger. The generated hypotheses such as AADP, AADDD, and DDRRNH, respectively, for ligand-, e-pharmacophore and receptor cavity based approach alongside shape of oseltamivir were successfully utilized to screen the DrugBank database. Subsequently, these models were evaluated for their differentiating ability using Enrichment calculation. Receiver operating curve and enrichment factors from the analysis indicate that the models possess better capability to screen actives from decoy set of molecules. Eventually, the hits retrieved from different hypotheses were subjected to molecular docking using Glide module of Schrödinger Suite. The results of different algorithms were then combined to eliminate false positive hits and to demonstrate reliable prediction performance than existing approaches. Of note, Pearson’s correlation coefficients were calculated to examine the extent of correlation between the glide score and IC50 values. Further, the interaction profile, pharmacokinetic, and pharmacodynamics properties were analyzed for the hit compounds. The results from our analysis showed that alprostadil (DB00770) exhibits better binding affinity toward NA protein than the existing drug molecules. The biological activity of the hit was also predicted using PASS algorithm that renders the antiviral activity of the compound. Further, the results were validated using mutation analysis and molecular dynamic simulation studies. Indeed, this integrative filtering is able to exceed accuracy of other state-of-the-art methods for the drug discovery.

中文翻译：

---

利用统计框架进行药物利用的多维筛选策略-流感药物发现的新途径。

流感病毒以影响全球数十亿人的间歇性爆发而闻名。在实践中已经使用了几种神经氨酸酶抑制剂来克服这种情况。然而，新的抗性突变体的出现限制了其临床应用。近年来，药物再利用技术因其具有成本效益并减少了发现药物所花费的时间而备受关注。在这里，我们提出了多维再利用策略，该策略整合了配体，能量，受体腔和基于形状的药效团算法的结果，以有效地识别新型流感候选药物。药效基团假说是通过使用Schrödinger的PHASE模块生成的。对于配体-，分别生成了假设，例如AADP，AADDD和DDRRNH 基于电子药效团和受体腔的方法以及奥司他韦的形状已成功用于筛选DrugBank数据库。随后，使用富集计算评估这些模型的区分能力。接收器工作曲线和分析的富集因子表明，该模型具有更好的能力，可从分子诱饵组中筛选出活性物质。最终，使用SchrödingerSuite的Glide模块对从不同假设中检索到的命中进行分子对接。然后将不同算法的结果进行组合，以消除误报，并证明与现有方法相比，预测性能可靠。值得注意的是，计算了Pearson的相关系数以检查滑行得分与IC50值之间的相关程度。进一步，分析了这些命中化合物的相互作用，药代动力学和药效学性质。我们的分析结果表明，前列地尔（DB00770）对NA蛋白的结合亲和力比现有药物分子更好。还使用PASS算法预测了命中的生物活性，该算法可提供化合物的抗病毒活性。此外，使用突变分析和分子动力学模拟研究验证了结果。实际上，这种集成过滤能够超过其他最新的药物发现方法的准确性。我们的分析结果表明，前列地尔（DB00770）对NA蛋白的结合亲和力比现有药物分子更好。还使用PASS算法预测了命中的生物活性，该算法可提供化合物的抗病毒活性。此外，使用突变分析和分子动力学模拟研究验证了结果。实际上，这种集成过滤能够超过其他最新的药物发现方法的准确性。我们的分析结果表明，前列地尔（DB00770）对NA蛋白的结合亲和力比现有药物分子更好。还使用PASS算法预测了命中的生物活性，该算法可提供化合物的抗病毒活性。此外，使用突变分析和分子动力学模拟研究验证了结果。实际上，这种集成过滤能够超过其他最新的药物发现方法的准确性。