Structured illumination microscopy combined with machine learning enables the high throughput analysis and classification of virus structure
Structured illumination microscopy combined with machine learning enables the high throughput analysis and classification of virus structure
复制标题
DOI：
10.7554/elife.40183
复制
发表时间：
2018-12-13
期刊：
影响因子：
7.7
通讯作者：
Kaminski, Clemens F.
中科院分区：
文献类型：
--
作者：
Laine, Romain F.;Goodfellow, Gemma;Kaminski, Clemens F.
光学超分辨率显微镜技术实现了高空间分辨率的高分子特异性，并构成了研究病毒等超分子组装体结构的一套强大工具。在这里，我们报告了一种新方法，该方法将结构照明显微镜（SIM）与机器学习算法相结合，以高分辨率对大量生物制药病毒的结构进行成像和分类。该方法提供了有关病毒形态的信息，最终可以与功能性能联系起来。我们展示了用于溶瘤病毒疗法（新城疫病毒）和疫苗开发（流感）的病毒的方法。这种独特的工具能够以高通量快速评估病毒生产的质量，从而避免了复杂且耗时的传统批量测试方法的需要。我们表明，我们的方法也适用于直接来自生产线的汇集收获液的未纯化样品。
Optical super-resolution microscopy techniques enable high molecular specificity with high spatial resolution and constitute a set of powerful tools in the investigation of the structure of supramolecular assemblies such as viruses. Here, we report on a new methodology which combines Structured Illumination Microscopy (SIM) with machine learning algorithms to image and classify the structure of large populations of biopharmaceutical viruses with high resolution. The method offers information on virus morphology that can ultimately be linked with functional performance. We demonstrate the approach on viruses produced for oncolytic viriotherapy (Newcastle Disease Virus) and vaccine development (Influenza). This unique tool enables the rapid assessment of the quality of viral production with high throughput obviating the need for traditional batch testing methods which are complex and time consuming. We show that our method also works on non-purified samples from pooled harvest fluids directly from the production line.