Using machine learning and big data to explore the drug resistance landscape in HIV.
Using machine learning and big data to explore the drug resistance landscape in HIV.
复制标题
DOI：
10.1371/journal.pcbi.1008873
复制
发表时间：
2021-08
影响因子：
4.3
通讯作者：
UK HIV Drug Resistance Database
中科院分区：
文献类型：
--
作者：
Blassel L;Tostevin A;Villabona-Arenas CJ;Peeters M;Hué S;Gascuel O;UK HIV Drug Resistance Database
在治疗压力下，艾滋病毒出现耐药突变（DRM）。DRM通常传播给未接受过治疗的患者。揭示新DRM的标准方法是测试治疗患者和未治疗患者之间突变的显著频率差异。然而，我们会单独考虑每个突变，不能指望研究几个突变之间的相互作用。在这里，我们的目标是利用不断增长的高质量序列数据和机器学习方法来研究这种相互作用（即上位性），并试图找到新的DRM。我们使用所有观察到的突变作为二进制表示特征，在来自英国的大型HIV-1逆转录酶（RT）序列数据集（n = 55，000）上训练分类器来区分逆转录酶抑制剂（RTI）经历过的样本和RTI未经历过的样本。为了评估我们研究结果的稳健性，我们的分类器在来自英国和非洲的独立数据集上进行了评估。然后提取每个分类器的重要表示特征作为潜在的DRM。为了找到新的DRM，我们通过删除与已知DRM相关的特征或样本来重复此过程。当保持所有已知的电阻信号，我们检测到足够普遍的已知DRM，从而验证该方法。当去除与已知DRM相对应的特征时，我们的分类器保留了一些预测准确性，并确定了与抗性显著相关的六个新突变。这六个突变具有低遗传屏障，与已知的DRM相关，并且在空间上接近RT活性位点或调节结合口袋。当去除已知的DRM特征和包含至少一个已知DRM的序列时，我们的分类器失去了所有的预测准确性。这些结果可能表明，所有直接赋予耐药性的突变都已发现，我们新发现的DRM是辅助或补偿突变。此外，除了我们发现的关系的附属性质之外，我们没有发现任何进一步的、更微妙的上位性结合几个单独似乎不赋予任何抗性的突变的显著信号。几乎所有治疗HIV的药物都靶向逆转录酶（RT）和耐药突变（DRMs），在治疗压力下HIV出现耐药突变。耐药菌株可以传播，并限制了人群水平的治疗选择。传统上，多重统计检验用于通过比较经治疗群体和未经治疗群体的病毒序列来发现DRM。然而，使用这种方法，每个突变都是单独考虑的，我们不能希望揭示它们之间的任何相互作用（上位性）。在这里，我们使用机器学习来发现新的DRM并研究潜在的上位效应。我们将这种方法应用于一个非常大的英国数据集，包括155，000个RT序列。在不同的英国和非洲数据集上检查了结果的稳健性。发现了6个与耐药性相关的新突变，所有6个突变都具有较低的遗传屏障，并与已知的DRM高度相关。此外，所有这些突变都接近RT的活性位点或调节结合口袋。因此，它们是进一步湿实验的良好候选者，以确定它们在耐药性中的作用。重要的是，我们的研究结果表明，上位性似乎是有限的经典方案，其中主要的DRM赋予电阻和相关的突变调节电阻的强度和/或补偿的健身成本诱导的DRM。
Drug resistance mutations (DRMs) appear in HIV under treatment pressure. DRMs are commonly transmitted to naive patients. The standard approach to reveal new DRMs is to test for significant frequency differences of mutations between treated and naive patients. However, we then consider each mutation individually and cannot hope to study interactions between several mutations. Here, we aim to leverage the ever-growing quantity of high-quality sequence data and machine learning methods to study such interactions (i.e. epistasis), as well as try to find new DRMs. We trained classifiers to discriminate between Reverse Transcriptase Inhibitor (RTI)-experienced and RTI-naive samples on a large HIV-1 reverse transcriptase (RT) sequence dataset from the UK (n ≈ 55, 000), using all observed mutations as binary representation features. To assess the robustness of our findings, our classifiers were evaluated on independent data sets, both from the UK and Africa. Important representation features for each classifier were then extracted as potential DRMs. To find novel DRMs, we repeated this process by removing either features or samples associated to known DRMs. When keeping all known resistance signal, we detected sufficiently prevalent known DRMs, thus validating the approach. When removing features corresponding to known DRMs, our classifiers retained some prediction accuracy, and six new mutations significantly associated with resistance were identified. These six mutations have a low genetic barrier, are correlated to known DRMs, and are spatially close to either the RT active site or the regulatory binding pocket. When removing both known DRM features and sequences containing at least one known DRM, our classifiers lose all prediction accuracy. These results likely indicate that all mutations directly conferring resistance have been found, and that our newly discovered DRMs are accessory or compensatory mutations. Moreover, apart from the accessory nature of the relationships we found, we did not find any significant signal of further, more subtle epistasis combining several mutations which individually do not seem to confer any resistance. Almost all drugs to treat HIV target the Reverse Transcriptase (RT) and Drug resistance mutations (DRMs) appear in HIV under treatment pressure. Resistant strains can be transmitted and limit treatment options at the population level. Classically, multiple statistical testing is used to find DRMs, by comparing virus sequences of treated and naive populations. However, with this method, each mutation is considered individually and we cannot hope to reveal any interaction (epistasis) between them. Here, we used machine learning to discover new DRMs and study potential epistasis effects. We applied this approach to a very large UK dataset comprising ≈ 55, 000 RT sequences. Results robustness was checked on different UK and African datasets. Six new mutations associated to resistance were found. All six have a low genetic barrier and show high correlations with known DRMs. Moreover, all these mutations are close to either the active site or the regulatory binding pocket of RT. Thus, they are good candidates for further wet experiments to establish their role in drug resistance. Importantly, our results indicate that epistasis seems to be limited to the classical scheme where primary DRMs confer resistance and associated mutations modulate the strength of the resistance and/or compensate for the fitness cost induced by DRMs.
登录
查看更多内容
影响因子：
10.2
作者：
Barathmanikanth S;Kalishwaralal K;Sriram M;Pandian SR;Youn HS;Eom S;Gurunathan S
通讯作者：
Gurunathan S
影响因子：
5.8
作者：
Draghici, S;Potter, RB
通讯作者：
Potter, RB
影响因子：
3.8
作者：
Mourad, Raphael;Chevennet, Francois;Hue, Stephane
通讯作者：
Hue, Stephane
影响因子：
4.4
作者：
Mooney, Alyssa C.;Campbell, Chadwick K.;Lippman, Sheri A.
通讯作者：
Lippman, Sheri A.
影响因子：
3.7
作者：
Bennett DE;Camacho RJ;Otelea D;Kuritzkes DR;Fleury H;Kiuchi M;Heneine W;Kantor R;Jordan MR;Schapiro JM;Vandamme AM;Sandstrom P;Boucher CA;van de Vijver D;Rhee SY;Liu TF;Pillay D;Shafer RW
通讯作者：
Shafer RW