At present, heart failure (HF) treatment only targets the symptoms based on the left ventricle dysfunction severity; however, the lack of systemic ‘omics’ studies and available biological data to uncover the heterogeneous underlying mechanisms signifies the need to shift the analytical paradigm towards network-centric and data mining approaches. This study, for the first time, aimed to investigate how bulk and single cell RNA-sequencing as well as the proteomics analysis of the human heart tissue can be integrated to uncover HF-specific networks and potential therapeutic targets or biomarkers. We also aimed to address the issue of dealing with a limited number of samples and to show how appropriate statistical models, enrichment with other datasets as well as machine learning-guided analysis can aid in such cases. Furthermore, we elucidated specific gene expression profiles using transcriptomic and mined data from public databases. This was achieved using the two-step machine learning algorithm to predict the likelihood of the therapeutic target or biomarker tractability based on a novel scoring system, which has also been introduced in this study. The described methodology could be very useful for the target or biomarker selection and evaluation during the pre-clinical therapeutics development stage as well as disease progression monitoring. In addition, the present study sheds new light into the complex aetiology of HF, differentiating between subtle changes in dilated cardiomyopathies (DCs) and ischemic cardiomyopathies (ICs) on the single cell, proteome and whole transcriptome level, demonstrating that HF might be dependent on the involvement of not only the cardiomyocytes but also on other cell populations. Identified tissue remodelling and inflammatory processes can be beneficial when selecting targeted pharmacological management for DCs or ICs, respectively.

中文翻译：

---

使用综合多组学方法对扩张型和缺血性心肌病的治疗靶点和生物标志物的洞察

目前，心力衰竭（HF）治疗仅针对左心室功能障碍严重程度的症状；然而，缺乏系统的“组学”研究和可用的生物学数据来揭示异质的潜在机制意味着需要将分析范式转向以网络为中心的数据挖掘方法。这项研究首次旨在研究如何整合大量和单细胞 RNA 测序以及人类心脏组织的蛋白质组学分析，以揭示 HF 特异性网络和潜在的治疗靶点或生物标志物。我们还旨在解决处理数量有限的样本的问题，并展示适当的统计模型、其他数据集的丰富性以及机器学习引导的分析如何在这种情况下提供帮助。此外，我们使用来自公共数据库的转录组和挖掘数据阐明了特定的基因表达谱。这是使用两步机器学习算法来实现的，该算法基于一种新的评分系统来预测治疗目标的可能性或生物标志物的易处理性，该系统也已在本研究中引入。所描述的方法对于临床前治疗开发阶段的目标或生物标志物选择和评估以及疾病进展监测可能非常有用。此外，本研究揭示了 HF 的复杂病因学，区分了单细胞、蛋白质组和全转录组水平上扩张型心肌病 (DCs) 和缺血性心肌病 (ICs) 的细微变化，证明 HF 可能不仅取决于心肌细胞的参与，还取决于其他细胞群的参与。分别为 DC 或 IC 选择靶向药理学管理时，已确定的组织重塑和炎症过程可能是有益的。