Data-driven inference of regulators for cytokine-mediated tumor killing

目的1、通过资料综合分析发现促进肿瘤进展的新细胞因子。我们将全面整合来自癌症免疫治疗研究的临床基因组数据，以确定新的可溶性蛋白，这些蛋白的基因表达水平与许多队列中的患者临床结果显著相关。然后，我们将利用小鼠肿瘤中的基因过度表达来验证排名靠前的候选基因对肿瘤进展的体内影响。对于有效的分泌蛋白，我们将生成治疗反应简档，以丰富CytoSig框架。目的2.提出一种识别ST段数据中分泌蛋白信号的算法。有许多计算方法可以从批量或单细胞转录数据中研究配体-受体的相互作用。然而，大多数分泌蛋白的受体(估计为1903)尚不清楚。此外，许多非受体蛋白可能是分泌蛋白功能的基本调节器或指示物。为了研究广泛的分泌蛋白的信号活性，我们将开发空间模式检测算法来识别其表达模式与分泌蛋白编码基因具有正或负空间相关性的功能相关基因。然后，将使用在AIM 1中建立的体内模型来验证最高预测的调控因子。

Aim 1, Discover new cytokines promoting tumor progression through data-integrative analysis. We will comprehensively integrate clinical genomics data from cancer immunotherapy studies to identify new soluble proteins whose gene expression levels are significantly associated with patient clinical outcomes in many cohorts. Then, we will validate the top candidates' in-vivo impact on tumor progression using gene over-expression in mouse tumors. For validated secreted proteins, we will generate treatment response profiles to enrich the CytoSig framework. Aim 2. Develop an algorithm to identify secreted protein signaling in ST data. Many computational methods exist for studying ligand-receptor interactions from bulk or single-cell transcriptomics data. However, the receptors for most secreted proteins (1903 by estimation) are unknown. Also, many non-receptor proteins may serve the essential regulators or indicators of secreted protein functions. To study signaling activities for a broad set of secreted proteins, we will develop spatial pattern detection algorithms to identify functional-relevant genes whose expression patterns have positive or negative spatial correlations with the coding gene of secreted proteins. Then, top predicted regulators will be validated using the in-vivo models established in Aim 1.