| Objective:CABLES1,as a substrate and anchorage protein for cdk5 and c-Abl,has been shown to have low expression level in tumor tissues and tumor suppressor effects in human non-small cell lung cancer[1],clone cancer[2,3],endometrial cancers[4,5]and ovarian cancer[6,7].but whether it has a similar effect in many tumors.The aim of this study was to investigate whether CABLES1 mRNA expression,copy number,methylation and mutation have altered in many tumors and their impact on clinical patient survival and prognosis,and whether CABLES1 copy number and methylation alternation can affect its mRNA expression in tumors.The mRNA expression,copy number,and methylation these multi-omics feature of CABLES1 if can use machine learning to establish a predictive model to predict the survival time of patients.Method:The RNAseq transcriptome data,copy number data,methylation data,mutation data and clinical data of 33 tumors were downloaded from the TCGA database using the Linux command line and gdc-client.The mRNA expression level of CBALES1 in 33 kinds of tumors was analyzed by R language,and the differential expression analysis of CABLES1mRNA expression in tumor tissues and normal tissues was performed using the edgeR package.Differential expression analysis of CABLES1 mRNA expression in different tumor stages and different clinical states was also analyzed.Survival analysis of high and low level mRNA expression of CBALES1 in tumor patients was performed using the survival package.The copy number level of CABLES1 and the survival analyze of copy number alternation of CABLES1 in 33 tumors was also analyzed,and the survival package was used for copy number amplification of CBALES1,and the survival analysis using R programming language.The correlation of between CABLES1 copy number alternation and its mRNA expression in33 cancers were analyzed using corr function of R.The CBALES1 methylation data were processed and analyzed by R and python language.The different methylation position between tumor tissues and normal tissues of CABLES1 was analyzed using limma package.The different methylation regions between tumor tissues and normal tissues of CABLES1were analyzed using DMRcate package.The correlation between CABLES1 methylation level and CABLES1 mRNA expression level was analyzed by corr function in R.The 33cancers mutation type and mutation frequency of CABLES1 were analyzed by genvisR.The cancer patient survival time prediction models were build using Adaboost Regression Tree,Random Forest Regreeion Tree,Gradient Boost Regression Tree from python sklearn library and CABLES1 mRNA expression,copy number variations and methylation data.Results:1.CABLES1 mRNA expression level was lower in tumor tissues than in normal tissues in GBM,LUSC,PCPG,HNSC,LUAD,KICH,THCA,THCA,BLCA,CESC,BRCA,KIRC,KIRP,READ,CHOL,ESCA.CABLES1 lower mRNA expression cancer patients have bad clinical prognosis in HNSC.CABLES1 have a lower mRNA expression in HNSC progressive disease patients and in perineural invasion patients.CABLES1 have a higher mRNA expression in HNSC complete remission or response patients.2.CABLES1 have copy number deletion in READ,TCTG,COAD,ACC,SKCM,LGG,KIRC,BRCA,PRAD,LIHC,KIRP.There are bad clinical prognosis and lower survival time in CABLES1 copy number deletion patients in BRCA and KIRP.There are high correlation between copy number and mRNA expression in READ,COAD,OV,HNSC,ESCA,STAD,UVM,LUAD,SKCM,PAAD,KIRP,UCS,KIRC,BRCA.3.CABLES1 DNA has lower methylation level in most TCGA cancers.CABLES1 have higher methylation level in its TSS upstream in LGG,ACC,GBM.CABLES1 also have higher methylation level in its TSS upstream in LGG,ACC.There are strong negative correlation between CABLES1 mRNA expression and CABLES1 TSS downstream methylation in PAAD,READ,SARC,UCS,ACC,BRCA,CESC,COAD,KICH,KIRC,KIRP,LAML.There are strong negative correlation between CABLES1 mRNA expression and CABLES1 TSS upstream（0-5kb）methylation in PAAD,PCPG,SARC,THYM,UCS,ACC,CESC,ESCA,KIRC,KIRP,LAML,LGG,LIHC,LUAD.4.CABLES1 have low mutation frequency in most cancer but have higher mutation in COAD,UCEC,MESO,BLCA,KIRP.The most mutation types of CABLES1 is missense mutation.Other high frequency mutation type are 3’-UTR,silent.These high frequency mutation type can have some influence to CBALES1 transcribe and protein expression.CABLES1 also have synonymous mutation in UCEC,COAD,BLCA,SKCM,KIRP,STAD,LGG,GBM.5.We use Adaboost Regression tree machine learning model with CABLES1 mRNA expression,copy number level and methylation level data can predict 90.9%COAD patient survival time very well.Conclusions:There are worse influence to man cancer patients’survival time and clinical prognosis because of low CABLES1 mRNA expression level,low copy number of CABLES1 and high methylation level these features.In many cancers we analyzed,CABLES1 methylation levels have strong negative correlation with CABLES1 mRNA expression levels in many cancers.CABLES1 copy number has strong positive correlation with CABLES1 mRNA expression level in many cancers.We use Adaboost Regression tree machine learning model with CABLES1 mRNA expression,copy number level and methylation level data can predict90.9%COAD patient survival time very well. |
