arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 1506.06392q-bio.MNq-bio.QM
收藏

生物多尺度过程的数据驱动建模

Data-driven modelling of biological multi-scale processes

Jan Hasenauer, Nick Jagiella, Sabrina Hross, Fabian J. Theis

摘要:

生物过程涉及各种空间和时间尺度。因此,对许多生物过程的整体理解需要多尺度模型,这些模型可以捕获所有这些尺度上的相关特性。在这篇手稿中,我们回顾数学建模方法用于描述个人的空间尺度,以及如何将它们集成到整体模型。我们讨论了空间尺度和时间尺度之间的关系,并对多尺度建模的含义。基于对最先进的建模方法的概述,我们制定了生物多尺度和多物理过程的数学和计算建模的关键挑战。特别是,我们考虑了多尺度模型和基于模型的多尺度数据集成的分析工具的可用性。我们提供了一个紧凑的审查基于模型的数据集成和基于模型的假设检验的方法。此外,新的方法和最近的趋势进行了讨论,包括计算时间减少使用降阶和代理模型,这有助于解决推理问题。最后,我们提供了一些想法的发展定制的多尺度推理方法的手稿。

英文摘要:

Biological processes involve a variety of spatial and temporal scales. A holistic understanding of many biological processes therefore requires multi-scale models which capture the relevant properties on all these scales. In this manuscript we review mathematical modelling approaches used to describe the individual spatial scales and how they are integrated into holistic models. We discuss the relation between spatial and temporal scales and the implication of that on multi-scale modelling. Based upon this overview over state-of-the-art modelling approaches, we formulate key challenges in mathematical and computational modelling of biological multi-scale and multi-physics processes. In particular, we considered the availability of analysis tools for multi-scale models and model-based multi-scale data integration. We provide a compact review of methods for model-based data integration and model-based hypothesis testing. Furthermore, novel approaches and recent trends are discussed, including computation time reduction using reduced order and surrogate models, which contribute to the solution of inference problems. We conclude the manuscript by providing a few ideas for the development of tailored multi-scale inference methods.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑