Molecular signatures of synergy

针对单个分子靶点的药物通常不足以对抗真菌感染、多基因疾病（如癌症）和多种细胞或组织类型疾病（包括免疫和炎症性疾病）。同时影响多个靶点的多组分治疗不容易产生耐药性，并提高治疗效果。多组分治疗的主要好处之一是潜在的协同效应：也就是说，药物组合的整体治疗效益大于单个药物的效果总和。特别是，成分化合物之间的协同作用可以提供更广泛的药理学窗口和降低毒性。有许多挑战与分子协同作用的识别有关，而且这种协同作用似乎很少见。事实上，即使是经过充分研究的酵母，也只有大约20种协同作用是已知的。高通量筛选可以检查数千个化合物对，但这只代表了可用化学组合指数数量的很小一部分。如果我们考虑到筛选文库中有1000万种化合物，那么实验技术不太可能足以以具有成本效益和及时的方式完全调查仅成对协同作用（~1013对）的广阔空间。显然需要一种方法，将这一空间筛选为可管理的大量组合，以丰富协同组合。我们建议构建基因组学和生物信息学相结合的方法来识别和表征分子协同作用。利用受到单一化学干扰的酵母或人类细胞系产生的高通量基因表达谱，我们的新生物信息学工具将预测极有可能表现出成对协同效应的化合物对。此外，分析由这对药物引起的下游转录变化，无论是联合还是单独，将提供对协同作用的分子性质的基本见解。

Drugs that act against individual molecular targets are often insufficient to combat fungal infections, multigenic diseases such as cancer, and multiple cell or tissue type diseases including immune and inflammatory disorders. Multicomponent therapies that impact several targets simultaneously are less prone to the development of drug resistance, and increase therapeutic efficacy. One of the major benefits of multicomponent therapies is the potential for synergistic effects: that is, the overall therapeutic benefit of the drug combination is greater than the sum of the effects of the individual agents. In particular, synergies between the constituent compounds can provide broader pharmacological windows and reduced toxicity. There are many challenges associated with the identification of molecular synergy and it appears that such synergies are rare. In fact, even for the well-studied yeast, only ~20 synergies are known. High-throughput screens can examine on the order of several thousand of compound pairs, but this represents a very small fraction of the exponential number of chemical combinations available. If we consider that there are >10 million compounds in screening libraries, it is unlikely that experimental techniques will be sufficient to completely survey the vast space of only pairwise synergies (~1013 pairs) in a cost effective and timely fashion. There is a clear need for an approach that winnows this space to a manageably large set of combinations that is enriched for synergistic combinations. We propose to construct a combined genomics and bioinformatics approach to identify and characterize molecular synergies. Using high-throughput gene expression profiles generated from yeast or human cell lines subjected to a single chemical perturbation, our de novo bioinformatics tool will predict pairs of chemical compounds which are highly likely to exhibit a pair-wise synergistic affect. Moreover, analysis of the downstream transcriptional changes caused by the pair of drugs, either in combination or individually, will provide fundamental insight into the molecular nature of synergy.