Supporting data for "Improved integration of single cell transcriptome data demonstrates common and unique signatures of heart failure in mice and humans"

Cardiovascular research heavily relies on mouse (<i>Mus musculus</i>) models to study disease mechanisms and to test novel biomarkers and medications. Yet, applying these results to patients remains a major challenge and often results in non-effective drugs. Therefore, it is an open challenge of translational science to develop models with high similarities and predictive value. This requires a comparison of disease models in mice with diseased tissue derived from humans. <br>To compare the transcriptional signatures at single cell resolution, we implemented an integration pipeline called <i>OrthoIntegrate</i> which uniquely assigns orthologs and therewith merges single cell data (scRNA-SEQ) of different species. The pipeline has been designed to be as easy to use and is fully integrable in the standard Seurat workflow. <br>We applied <i>OrthoIntegrate</i> on scRNA-SEQ from cardiac tissue of heart failure patients with reduced ejection fraction (HFrEF) and scRNA-SEQ from the mice after chronic infarction, which is a commonly used mouse model to mimic HFrEF. We discovered shared and distinct regulatory pathways between human HFrEF patients and the corresponding mouse model. Overall, 54% of genes were commonly regulated including major changes in cardiomyocyte energy metabolism. However, several regulatory pathways, e.g. angiogenesis, were specifically regulated in humans. <br>The demonstration of unique pathways occurring in humans indicate limitations on the comparability between mice models and human HFrEF and show that results from the mice model should be validated carefully. <i>OrthoIntegrate</i> is publicly accessible (https://github.com/MarianoRuzJurado/OrthoIntegrate) and can be used to integrate other large data sets to provide a general comparison of models with patients data.

心血管研究高度依赖小鼠（*Mus musculus*）模型，用于探究疾病发病机制、筛选新型生物标志物与候选药物。然而，将此类研究成果转化应用于临床患者仍面临巨大挑战，且常导致药物研发失败。因此，开发具有高度相似性与预测价值的疾病模型，是转化医学领域亟待解决的开放性难题。这一目标需要开展小鼠疾病模型与人类病变组织的比对分析。

为实现单细胞分辨率下的转录组特征比对，我们搭建了一款名为OrthoIntegrate的整合分析流程，该流程通过唯一同源基因分配策略，可整合不同物种的单细胞RNA测序（scRNA-SEQ）数据。本流程旨在简化操作，且可完全兼容标准Seurat工作流。

我们将OrthoIntegrate应用于射血分数降低型心力衰竭（heart failure with reduced ejection fraction, HFrEF）患者心脏组织的scRNA-SEQ数据，以及模拟HFrEF的经典慢性梗死小鼠模型的scRNA-SEQ数据。研究发现，人类HFrEF患者与对应小鼠模型之间既存在共有调控通路，也存在特异性调控通路。整体而言，54%的基因存在共同调控模式，其中包括心肌细胞能量代谢的显著变化。但部分调控通路（如血管生成通路）仅在人类样本中呈现特异性调控特征。

本研究证实人类样本中存在特异性调控通路，表明小鼠模型与人类HFrEF之间的可比性存在局限，同时提示需谨慎验证小鼠模型的研究结果。OrthoIntegrate已公开部署（https://github.com/MarianoRuzJurado/OrthoIntegrate），可用于整合其他大型数据集，以实现各类疾病模型与患者数据的通用比对分析。