Feature weighted models to address lineage dependency in drug-resistance prediction from Mycobacterium tuberculosis genome sequences.
Feature weighted models to address lineage dependency in drug-resistance prediction from Mycobacterium tuberculosis genome sequences.
复制标题
DOI：
10.1093/bioinformatics/btad428
复制
发表时间：
2023-07-01
期刊：
影响因子：
--
通讯作者：
中科院分区：
文献类型：
--
作者：
结核病 (TB) 是由结核分枝杆菌复合体 (MTBC) 的成员引起的，该复合体具有基于菌株或谱系的克隆群体结构。 MTBC 耐药性的演变对结核病的成功治疗和根除构成威胁。机器学习方法越来越多地被用来预测耐药性并表征全基因组序列的潜在突变。然而，由于 MTBC 群体结构的混淆，这种方法可能无法很好地推广到临床实践中。为了研究群体结构如何影响机器学习预测，我们比较了三种不同的方法来减少随机森林（RF）模型中的谱系依赖，包括分层、特征选择和特征加权模型。所有 RF 模型都实现了中高性能（ROC 曲线范围下的面积：0.60–0.98）。一线药物的性能高于二线药物，但其性能因训练数据集中的谱系而异。谱系特异性模型通常比全局模型具有更高的敏感性，这可能受到菌株特异性耐药突变或采样效应的支持。特征权重和特征选择方法的应用减少了模型中的谱系依赖性，并且具有与未加权 RF 模型相当的性能。 https://github.com/NinaMercedes/RF_lineages。
Tuberculosis (TB) is caused by members of the Mycobacterium tuberculosis complex (MTBC), which has a strain- or lineage-based clonal population structure. The evolution of drug-resistance in the MTBC poses a threat to successful treatment and eradication of TB. Machine learning approaches are being increasingly adopted to predict drug-resistance and characterize underlying mutations from whole genome sequences. However, such approaches may not generalize well in clinical practice due to confounding from the population structure of the MTBC. To investigate how population structure affects machine learning prediction, we compared three different approaches to reduce lineage dependency in random forest (RF) models, including stratification, feature selection, and feature weighted models. All RF models achieved moderate-high performance (area under the ROC curve range: 0.60–0.98). First-line drugs had higher performance than second-line drugs, but it varied depending on the lineages in the training dataset. Lineage-specific models generally had higher sensitivity than global models which may be underpinned by strain-specific drug-resistance mutations or sampling effects. The application of feature weights and feature selection approaches reduced lineage dependency in the model and had comparable performance to unweighted RF models. https://github.com/NinaMercedes/RF_lineages.
登录
查看更多内容
影响因子：
6
作者：
Huang, Chenxi;Zhou, Junsheng;Peng, Yonghong
通讯作者：
Peng, Yonghong
影响因子：
4.3
作者：
Collins C;Didelot X
通讯作者：
Didelot X
影响因子：
5.2
作者：
Kouchaki, Samaneh;Yang, Yang;Clifton, David A.
通讯作者：
Clifton, David A.
影响因子：
4.4
作者：
Deelder W;Napier G;Campino S;Palla L;Phelan J;Clark TG
通讯作者：
Clark TG
影响因子：
4.9
作者：
de Vos, M.;Mueller, B.;Victor, T. C.
通讯作者：
Victor, T. C.