Pan-cancer Biomarker Identification For Diagnosis From Specifically Expressed Genes

Cancer is the most deadly disease,which seriously threatens human health.Although there are great progresses,cancer is still a serious threat to human health because of the lack of biomarkers for diagnosis,especially early-stage diagnosis.Luckily,data platforms such as TCGA and GTEx provide a huge amount of public data which allow us to screen cancer tissue specific genes among multiple cancers.In this study,we comprehensively surveyed the specifically expressed genes(SEGs)by SEGtool based on the expression big data from TCGA and GTEx.As a result in 15 solid tumors,we identified 283 cancer SEGs(cSEGs),which were specifically expressed in only one cancer and had great potential to be biomarkers.Among them,three cSEGs(OGDH,MUDENG and ACO2)occupied sample frequency >80% in kidney cancer,suggesting their high sensitivity.Meanwhile,we identified 245 cSEGs as early-stage diagnostic biomarkers across 13 cancers.Among them,Two early stage cSEGs(MUDENG and ACO2)occupied sample frequency >95% in kidney cancer.A double-gene combination strategy was used to improve the sensitivity of diagnostic biomarkers and hundreds of double-gene combinations were identified with high frequency.The highest double-gene highly expressed sample frequency was 97% in KICH,and 100% in early stage KICH and COAD.We also identified 15 SEGs were targets of various anticancer drugs and about half of these drugs may be repurposed to cancers with their targets as SEGs.Some SEGs were regulated by the specific transcription factors in the corresponding cancer and 39 cSEGs were prognosis related genes in 7 cancers.This work provides a survey of cancer biomarkers for diagnosis and early diagnosis,as well as new insights to drug repurposing.These may have great potential and significance in cancer research and application.