ABSTRACT:"Big data," a consequence of the "omics" technologies and its analysis by machine learning, have changed the climate of thought in biomedical sciences, shifting the demography of expertise and culminating in a new role: "data scientist." While historically the inquiry on the nature of organisms started with theories (logical reasoning) but no data, we now live in an era of data but no theory. A tacit assumption of modern data analytics is that correlations and clusters in the data constitute knowledge. Through support of technology and data collection, funding agencies promoted this attitude, while neglecting hypothesis-driven inquiry and theory. Data is, of course, an indispensable ingredient of knowledge, but it cannot be the endpoint of inquiry. This article provides key concepts for a fruitful discussion, examines the dualism between data and theory, and proposes how they synergize. Data scientists must learn to appreciate theory, but if the most value is to be extracted from data, theorists should not dismiss brute-force empirical pattern recognition in data. The patterns could motivate the erection of new theories, much as Kepler's law represented a formal "summary" of astronomic data on which Newton's laws could be tested.

中文翻译：

---

生物医学研究“组学”时代大数据与理论的张力

摘要：“大数据”是“组学”技术及其机器学习分析的结果，改变了生物医学科学的思想氛围，改变了专业知识的人口结构，并最终形成了一个新角色：“数据科学家”。虽然从历史上看，对有机体本质的探究始于理论（逻辑推理）但没有数据，但我们现在生活在一个有数据但没有理论的时代。现代数据分析的一个默认假设是数据中的相关性和集群构成了知识。通过技术和数据收集的支持，资助机构促进了这种态度，而忽视了假设驱动的调查和理论。数据当然是知识不可或缺的组成部分，但它不能成为探究的终点。本文提供了富有成效的讨论的关键概念，检查数据和理论之间的二元论，并提出它们如何协同作用。数据科学家必须学会欣赏理论，但如果要从数据中提取最大价值，理论家就不应该忽视数据中的蛮力经验模式识别。这些模式可以激发新理论的建立，就像开普勒定律代表了可以测试牛顿定律的天文数据的正式“摘要”一样。