Identification Of Melanoma Driver Mutation And Construction Of Related Information Platform

Melanoma,also known as malignant melanoma,is a type of cancer that develops from melanocytes.It is usually caused by DNA damage through exposure to sunlight and different genetic alterations are also associated with melanoma.With the development of human genome sequencing technology,the melanoma genome atlas has been understood.The synergistic effect of mutations and other genetic mutation changes in cancer cells plays a role in signal transduction pathways associated with cancer cell growth and cell proliferation.By identifying key pathways that are active in melanoma cells,researchers can identify driver gene in these pathways,which can avail to targeted therapy for disease.In this paper,based on sequencing data and other omics data,we focus on database construction and the research on gene and mutation level of melanoma.With the development of genomic technology,more and more melanoma related genes have been reported.However,these genes are scattered and there are no database of melanoma genes,to solve this,we construct melanoma gene database.At the genetic level,we design approaches to identify driver genes.In order to more accurately predict driver mutations,we construct a model for predicting missense mutations in melanoma.Finally,as the mechanism of synonymous mutations in diseases has been revealed,we further explore the possible mechanism of synonymous mutations in melanoma.The work of this article is summarized as follows:1.Construction of melanoma gene database.The Melanoma Gene Database(MGDB)is a manually curated catalog of molecular genetic data relating to genes involved in melanoma.The main purpose of this database is to establish a network of melanoma related genes and to facilitate the mechanistic study of melanoma tumorigenesis.Through literature reading and database exploration,527 human melanoma genes were in MGDB(422 protein-coding and 105 non-coding genes).Each melanoma gene was annotated in seven different aspects(General Information,Expression,Methylation,Mutation,Interaction,Pathway,and Drug).In addition,manually curated literature references have also been provided to support the inclusion of the gene in MGDB and establish its association with melanoma.MGDB has a user-friendly web interface with multiple browse and search functions.Besides,in order to assess these melanoma genes,we performed network analysis and bio-functional enrichment assays and found that these genes are indeed involved in the development of melanoma.2.Prediction algorithm of melanoma driver genes.According to the mutation data and the protein interaction data,each mutant gene is firstly mapped to the protein interaction network.Then a neighbor network is constructed for each mutant gene in the interaction network.The significant neighbor network is found by a statistical method.Combined with the mutation frequency and the significance of mutant gene in significant neighbor network,each mutant gene was scored.In addition,we use the multi-omic data to detect driver genes.Based on the multi-omic data in melanoma,a similarity matrix between patients and patients was constructed based on the similarity matrix fusion algorithm.Gene and gene similarity were constructed based on gene annotation information.According to the mutation information of the sample,the relationship between the gene and the sample is constructed.Finally,the driver genes are ordered using sparse subspaces.The first method we built is more accurate than other algorithm on the same benchmark dataset.Functional and pathway enrichment analysis also shows that these driver genes are critical for melanoma.3.The prediction model construction of melanoma missense mutations.In this work,twenty features(including conservation score,pfam domain,functional region annotation,distance to nearest splice site,protein structure and physicochemical change of amino acid)were used to construct the model.In the training set,we obtained an AUC of 94%with 10 fold cross validation using random forest.The method was tested on test dataset,which showed an AUC of 94%.Compared with other tools,the main advantage of our method is that there is no bias in predicting the positive and negative samples.Most of these features are statistically significant in positive and negative samples,and we can infer that these features play an important role in predicting deleterious missense mutations in melanoma.4.The research on mechanism of synonymous mutations in melanoma.Though synonymous mutations do not change the protein sequence,cumulative evidence shows that they may affect protein function through splicing and other mechanisms.However,synonymous mutations in melanoma genomic field are rarely investigated.In this study,we analyzed pathogenic synonymous mutations in TCGA(The Cancer Genome Atlas)melanoma samples.Concentrating on five types of genetic regulatory function,we found that the synonymous mutations contribute to melanoma by three mechanisms:exonic splicing regulators in near-splicing site or locates in DNase I hypersensitivity sites or uses non-optimal codon.What’s more,the site of miRNA binding alteration exhibits a lower rate of evolution than other sites.We found 12 genes are hit by recurrent functional synonymous mutations.Among them,GRIN2A and TECTA genes have been reported to be mutated in melanoma.The rest genes may have a great potential to impact melanoma.These findings indicate that the role of synonymous mutations in melanoma,it attracts attention to researches.