基于转录调控网络预测肺鳞癌风险通路

Prediction of lung squamous carcinoma risk pathways based on transcriptional regulation network

目的肺鳞癌又称肺鳞状细胞癌,是发病率和致死率非常高的一种恶性肿瘤。肺鳞癌的病因尚不明确,本研究结合肿瘤基因组图谱(the cancer genome atlas,TCGA)肺鳞癌样本转录组数据,构建转录调控网络,采用广度优先搜索算法,识别肺鳞癌风险通路,探究肺鳞癌的发病机制。方法整合8个公共数据库资源,得到转录因子(transcription factor,TF)、长链非编码RNA(long non-coding RNA,lncRNA)、微小RNA(microRNA,miRNA)和基因(Gene)候选调控关系数据。然后从TCGA数据库下载肺鳞癌样本TF、lncRNA、miRNA和Gene的表达谱数据,计算候选调控关系的斯皮尔曼相关系数,构建肺鳞癌相关的转录调控网络。采用广度优先搜索算法,以每个入度为0的点作为起始节点,识别调控网络中的级联调控通路。同时,从疾病数据库中下载肺癌相关的TF、lncRNA、miRNA和Gene,对得到的级联调控链进行疾病注释,疾病注释率高的调控链被认为是肺鳞癌风险通路。结果整合8个数据库候选调控关系,包括TF对lncRNA、miRNA、Gene调控,lncRNA对TF、Gene调控以及miRNA对TF、lncRNA、Gene调控,共得到21 597对调控关系。结合TCGA中肺鳞癌样本的表达数据,在P<0.01的阈值下,最终得到肺鳞癌相关转录调控网络,包含2 234对调控关系,由91个TF、552个lncRNA、125个miRNA和937个Gene构成。采用广度优先搜索算法,共得到了10 941个调控级联链。经过疾病信息注释后,得到36条肺鳞癌注释率为1的肺鳞癌风险通路。最后,结合识别的风险通路和小分子药物与miRNA关系数据库,得到9种治疗肺鳞癌的候选药物。结论通过构建肺鳞癌相关转录调控网络,采用广度优先算法,可以识别肺鳞癌风险通路,为解释肺鳞癌的发病机制和预测治疗药物提供帮助。

OBJECTIVE Lung squamous carcinoma,also known as lung squamous cell carcinoma,is a very high incidence and mortality of malignant tumor.The pathogenesis of lung squamous carcinoma is not clear,and we here presented an effective method to identify lung squamous carcinoma risk pathway,helping to elucidate the pathogenesis of lung squamous carcinoma.METHODS We first integrated eight public database resources to obtain regulation data about transcription factor（TF）,long non-coding RNA（lncRNA）,miRNA and Gene.Then transcriptome sequencing data including TF,lncRNA,miRNA and Gene about lung squamous carcinoma were obtained from TCGA（The Cancer Genome Atlas）data resources.Spearman correlation coefficients were calculated about these regulation relationships,and we got a transcriptional regulatory network associated with lung squamous cell carcinoma.Taking each node with zero in-degree as the starting node,we identified the cascade pathways from the network using the breadth-first search algorithm.Lung cancer related TF,lncRNA,miRNA and Gene were downloaded from disease databases,and we annotated the cascade pathways with lung cancer information.The cascades with high disease annotation rate were regarded as lung squamous carcinoma risk pathways.RESULTS We integrated eight public database resources to retrieve 21 597 candidate regulation relationships,including TF versus lncRNA,miRNA and Gene,lncRNA versus TF and Gene and miRNA versus TF,lncRNA and Gene.After filtering the pairs with spearman correlation coefficients P-value0.01,we constructed a lung squamous carcinoma related regulatory network,including 2 234 pairs consisted of 91 TFs,552 lncRNAs,125 miRNAs and 937 Genes.By breadth-first search algorithm,we identified 10 941 cascades.After annotated these cascades with lung cancer information,36 cascades with annotation ratio 1 were obtained,and these cascades were deemed as lung squamous carcinoma risk pathways.In the end,we provided nine candidate small molecules for lung squamous carcinoma therapy based on identified risk pathways.CONCLUSION The constructed transcriptional regulation network can help to identify lung squamous carcinoma risk pathways,explain the pathogenesis of lung squamous cell carcinoma and predict the candidate clinical drugs for lung squamous carcinoma therapy.