不同基因的突变通常导致临床上相似的疾病。在类似疾病的数据集中,我们分析了来自人类在线孟德尔遗传的“表型系列”,并检查了属于同一表型系列的疾病的相似性,因为我们假设临床相似性可能揭示共同的致病机制。
具体来说,对于每一对疾病,我们量化了它们的相似性,基于共享临床表型的数量和信息内容。然后,我们组建了疾病相似网络,其中节点代表疾病,边缘代表临床相似性。
平均而言,疾病与自己的表型系列的其他疾病有很高的相似性,尽管大约三分之一的疾病与另一系列疾病有最大的相似性。因此,网络是分类的(即,属于同一系列的疾病优先彼此链接),但是该系列在网络中的分布方式有所不同。具体来说,疏异系列,最小化与其他系列的链接,在网络外围形成岛屿,而异亲性系列,与其他系列高度互连,占据网络的中心。
发现表型系列不仅显示内部相似性(同种性),而且显示不同程度的外部相似性(从异源性到异源性),这要求研究可能在不同系列之间共享的生物学机制。在本研究的第二部分1中分析了表型系列的临床和生物学相似性之间的相关性。
Mutations of different genes often result in clinically similar diseases. Among the datasets of similar diseases, we analyzed the \'phenotypic series\' from Online Mendelian Inheritance in Man and examined the similarity of the diseases that belong to the same phenotypic series, because we hypothesize that clinical similarity may unveil shared pathogenic mechanisms.
Specifically, for each pair of diseases, we quantified their similarity, based on both number and information content of the shared clinical phenotypes. Then, we assembled the disease similarity network, in which nodes represent diseases and edges represent clinical similarities.
On average, diseases have high similarity with other diseases of their own phenotypic series, even though about one third of diseases have their maximal similarity with a disease of another series. Consequently, the network is assortative (i.e., diseases belonging to the same series link preferentially to each other), but the series differ in the way they distribute within the network. Specifically, heterophobic series, which minimize links to other series, form islands at the periphery of the network, whereas heterophilic series, which are highly inter-connected with other series, occupy the center of the network.
The finding that the phenotypic series display not only internal similarity (assortativity) but also varying degrees of external similarity (ranging from heterophobicity to heterophilicity) calls for investigation of biological mechanisms that might be shared among different series. The correlation between the clinical and biological similarities of the phenotypic series is analyzed in Part II of this study1.