iSubGen 通过两两相似度评估生成整合疾病亚型。

iSubGen generates integrative disease subtypes by pairwise similarity assessment.

机构信息

Department of Medical Biophysics, University of Toronto, Toronto, ON M5G 1L7, Canada; Department of Human Genetics, University of California, Los Angeles, Los Angeles, CA, USA; Institute for Precision Health, University of California, Los Angeles, Los Angeles, CA, USA; Jonsson Comprehensive Cancer Center, University of California, Los Angeles, Los Angeles, CA, USA; Ontario Institute for Cancer Research, Toronto, ON M5G 0A3, Canada.

Department of Human Genetics, University of California, Los Angeles, Los Angeles, CA, USA; Institute for Precision Health, University of California, Los Angeles, Los Angeles, CA, USA; Jonsson Comprehensive Cancer Center, University of California, Los Angeles, Los Angeles, CA, USA.

出版信息

Cell Rep Methods. 2024 Nov 18;4(11):100884. doi: 10.1016/j.crmeth.2024.100884. Epub 2024 Oct 23.

DOI:10.1016/j.crmeth.2024.100884

PMID:39447572

原文链接:https://pmc.ncbi.nlm.nih.gov/articles/PMC11705582/

Abstract

There are myriad types of biomedical data-molecular, clinical images, and others. When a group of patients with the same underlying disease exhibits similarities across multiple types of data, this is called a subtype. Existing subtyping approaches struggle to handle diverse data types with missing information. To improve subtype discovery, we exploited changes in the correlation-structure between different data types to create iSubGen, an algorithm for integrative subtype generation. iSubGen can accommodate any feature that can be compared with a similarity metric to create subtypes versatilely. It can combine arbitrary data types for subtype discovery, such as merging genetic, transcriptomic, proteomic, and pathway data. iSubGen recapitulates known subtypes across multiple cancers even with substantial missing data and identifies subtypes with distinct clinical behaviors. It performs equally with or superior to other subtyping methods, offering greater stability and robustness to missing data and flexibility to new data types. It is available at https://cran.r-project.org/web/packages/iSubGen.

摘要

有无数种类型的生物医学数据——分子、临床图像等。当一群患有相同潜在疾病的患者在多种类型的数据中表现出相似性时，这就称为亚型(subtype)。现有的亚型发现方法难以处理具有缺失信息的多种数据类型。为了改善亚型发现，我们利用不同数据类型之间的相关性结构的变化，开发了一种名为 iSubGen 的综合亚型生成算法。iSubGen 可以适应任何可以用相似性度量进行比较的特征，从而灵活地创建各种类型的亚型。它可以结合任意数据类型进行亚型发现，例如合并遗传、转录组、蛋白质组和通路数据。iSubGen 即使在存在大量缺失数据的情况下，也可以重现多种癌症的已知亚型，并识别出具有不同临床行为的亚型(subtype)。它的性能与其他亚型发现方法相当或更优，对缺失数据具有更强的稳定性和鲁棒性，并且对新的数据类型具有更大的灵活性。它可在 https://cran.r-project.org/web/packages/iSubGen 上获取。