The Distribution Of Base Mutation In Cancer And Its Impact On The Mutational Landscape Of P53 And Other Genes

Cancer is a complex disease induced by many factors,of which genetic mutations are the main cause of tumorigenesis.At present,the main methods of treating cancer include surgical resection,chemoradiotherapy,small molecule targeted drugs,immunotherapy and so on.With the rapid development of big data analysis and deep-sequencing techonology,people have learnt more about the pathogenesis of cancer,which has greatly promoted the process of precision medicine in cancer.The complex landscape of cancer mutations in large database such as COSMIC renders it difficult to guage the functional status of these mutations.People generally predict the function of a certain mutation by its frequency in cancer genome database.If a mutation is detected in a large number of cancer samples,it's more likely assumed to be a driver mutation.Based on this method,people have found some important mutations(hotspot mutations)on major cancer genes,such as BRAF V600 E,p53R175H,etc.These mutations occur more frequently in various types of tumors and significantly promote the occurrence of cancer.However,such a simple method has obvious shortcomings.Most mutation in cancer genome database occur in very small number of samples,which make it difficult to predict their function.In order to solve this problem,we constructed a mathematical model trying to investigate the functionality of low-frequency cancer mutations.The possibility of observing a mutation in cancer samples is determined by at least two factors: first,whether the gene mutation is prone to occur at the base level;second,whether the gene mutation can promote the development of tumorigenesis.If a mutation can promotes tumorigenesis,but it is extremely difficult to generate such mutation at base level,the observed frequency of this mutation in cancer database will be very low.It is expected that factors including spontaneous deamination of nucleobase,DNA damage and repair mechanisms,as well as exposure rates to different carcinogens could all impact the mutation landscape in cancer.It is difficult to establish a model that could forwardly predict the relative chance of generating each type of mutation.Therefore,we catalogued more than three million of point mutations from 26154 cancer genomes in the COSMIC database,which allowed us to reversely extract the relative mutational difficulty for different types of nucleotide mutations.Our results show that the relative difficulty to generate different nucleotide mutations can vary by as much as 400-fold.Therefore,it is necessary to take both the observed mutation frequency and the difficulty to generate the underlying mutation into consideration when predicting the functionality of a certain mutation.Based on this,we constructed a new method which combined the original observed mutation frequency and the relative mutational difficulty to predict the functionality of cancer mutation.In order to test the accuracy of our method,we selected more than 80 low frequency mutations on p53 for experimental verification.The results showed that our method could accurately predict the functional status of these mutations.Furthermore,our method can also help discover new cancer-driving genes.It may also help understand the mutation process behind drug resistance in cancer.Our method provides a new angle for the better understanding of cancer etiology.In addition to predicting the functionality of individual cancer mutation,our model also provides new ideas for the prevention of cancer based on the perspective of relative mutational difficulty.