Identification of groups of co-expressed or co-regulated genes is critical for exploring the underlying mechanism behind a particular disease like cancer. Condition-specific (disease-specific) gene-expression profiles acquired from different platforms are widely utilized by researchers to get insight into the regulatory mechanism of the disease. Several clustering algorithms are developed using gene expression profiles to identify the group of similar genes. These algorithms are computationally efficient but are not able to capture the functional similarity present between the genes, which is very important from a biological perspective. In this study, an algorithm named CorGO is introduced, that specifically deals with the identification of functionally similar gene-clusters. Two types of relationships are calculated for this purpose. Firstly, the Correlation (Cor) between the genes are captured from the gene-expression data, which helps in deciphering the relationship between genes based on its expression across several diseased samples. Secondly, Gene Ontology (GO)-based semantic similarity information available for the genes is utilized, that helps in adding up biological relevance to the identified gene-clusters. A similarity measure is defined by integrating these two components that help in the identification of homogeneous and functionally similar groups of genes. CorGO is applied to four different types of gene expression profiles of different types of cancer. Gene-clusters identified by CorGO, are further validated by pathway enrichment, disease enrichment, and network analysis. These biological analyses demonstrated significant connectivity and functional relatedness within the genes of the same cluster. A comparative study with commonly used clustering algorithms is also performed to show the efficacy of the proposed method.

共表达或共调控基因组的鉴定对于探索特定疾病（如癌症）背后的潜在机制至关重要。从不同平台获得的条件特异性（疾病特异性）基因表达谱被研究人员广泛利用，以深入了解疾病的调控机制。使用基因表达谱开发了几种聚类算法来识别相似基因组。这些算法在计算上是高效的，但无法捕捉基因之间存在的功能相似性，这从生物学的角度来看非常重要。在这项研究中，引入了一种名为 CorGO 的算法，该算法专门处理功能相似的基因簇的识别。为此目的计算两种类型的关系。首先，基因之间的相关性 (Cor) 是从基因表达数据中捕获的，这有助于根据基因在多个患病样本中的表达来破译基因之间的关系。其次，利用可用于基因的基于基因本体论 (GO) 的语义相似性信息，这有助于将生物学相关性与已识别的基因簇相加。相似性度量是通过整合这两个组件来定义的，这两个组件有助于识别同质和功能相似的基因组。CorGO 应用于四种不同类型癌症的基因表达谱。CorGO 鉴定的基因簇通过通路富集、疾病富集和网络分析进一步验证。这些生物学分析证明了同一簇基因内的显着连通性和功能相关性。还进行了与常用聚类算法的比较研究，以显示所提出方法的有效性。