Multifactor Dimensionality Reduction Reveals The Effect Of Interaction Between Psoriasis Susceptibility Genes

Background:Psoriasis is a common immune-mediated skin proliferative dysfunction,and multiple related studies have reported the genetic susceptibility to psoriasis.To explore the effect of genetic variants on psoriasis,current studies use multiple methods to identify and detect psoriasis susceptibility genes,including genome-wide association studies（GWAS）,among others.Published studies have identified more than 50chromosomal regions associated with psoriasis susceptibility through GWAS,some of which contain multiple independent psoriasis susceptibility loci.However,most of the genetic factors of diseases cannot be explained by the independent action of a single gene,and the interaction between multiple genes is equally worthy of our attention for the genetic specificity of diseases.Given the multifactorial nature of chronic disease,risk assessment using a single candidate gene variant may not reveal a real impact on disease outcome.Non-Mendelian forms of inheritance,prevalent features of biomolecular communication,and inconsistent results from several single-variant,single-gene studies all support the idea that gene-gene interactions are common and can best describe disease symptoms.Currently,gene-gene interactions have been reported in many different disease populations.Multiple methods can be used to detect significant associations of gene-gene interactions with disease phenotypes.These statistical methods include logistic regression,neural networks,and multifactorial dimensionality reduction（MDR）.Among them,multifactorial dimensionality reduction is one of the most widely used methods.In this study,we performed a meta-analysis of psoriasis susceptibility loci in datasets from exome sequencing,target sequencing,and validation analysis by screening previous GWAS research databases and validating some psoriasis susceptibility loci in independent cohorts of psoriasis in order to expand the sample size and more effectively validate psoriasis susceptibility genes.We used validated genetic loci and statistically significant loci after meta-analysis to perform interaction analysis by multivariate dimensionality reduction to better validate the important role of these genes in psoriasis genetic specificity.Objective:In this study,we validated psoriasis susceptibility loci based on a database of previous studies and investigated interactions between validated psoriasis susceptibility loci.Methods:First,we validated psoriasis susceptibility gene variants in an independent cohort of 5,414 psoriasis patients and 5,556 controls.Multivariate dimensionality reduction（MDR）analysis was performed between loci that were statistically significant in the validation results to identify interactions between loci significantly associated with psoriasis in the validation cohort.We then performed a meta-analysis of these variants using datasets from exome sequencing,target sequencing,and validation analyses,and used MDR assays to identify the best gene-gene interaction model.Results:First,we screened 45 selected loci for validation in exome sequencing,target sequencing from previous studies,and 25 of these 45 psoriasis-associated loci were statistically significant in the validation analysis of an independent validation cohort（p<0.05）.Among them,ERAP1 was the most strongly associated with psoriasis(chr596118852,P=9.92×10^-9,OR=0.85),and other validated psoriasis susceptibility genes were IFIH1（2q24.3）,ERAP2（5q15）,IL18R1（2q12.1）,LTB（12p13.31）,IL1RL1（2q12.1）,CARD14（17q25.3）and SLC9A4（2q12.1）.Among these 25 psoriasis susceptibility SNPs,we performed an interaction analysis using the MDR method and showed that the rs27044 polymorphism of ERAP1 was the best unit-site model（cross-validation concordance=10,test precision=0.52,OR=1.32（95%CI,1.07 to1.63）,P<0.01）.The best two-locus prediction model was found between ERAP1rs27044 polymorphism and IFIH1 rs7590692 polymorphism（cross-validation concordance=9/10,test precision=0.53,OR=1.32（95%CI,1.09-1.59）,P<0.01）.In order to improve the validation accuracy and expand the sample size,we performed a meta-analysis of 45 psoriasis-associated SNPs selected by METAL software in exome sequencing,target sequencing,and validation cohorts,and the meta-analysis results showed that 12 of these SNPs reached genome-wide significance(P<10^-8).These SNPs were located in ERAP1(rs27044,P=4.05×10^-20,rs30378,P=3.41×10^-19,rs26510,P=1.94×10-18,rs27043,P=9.23×10^-18,rs469758,P=6.55×10-¹⁷,rs30186,P=3.02×10^-15),IFIH1(rs12479043,P=2.85×10^-15;rs7590692,P=1.57×10-¹⁴),CARD14(rs4889997,P=4.03×10^-12),GJB2(rs72474224,P=8.22×10^-11),NFKB1(rs3817685,P=4.24×10^-10),and ERAP2(rs2303208,P=7.50×10^-10),and again interaction analysis of these loci using MDR software showed that the best interaction model remained between the ERAP1 rs27044 polymorphism and the IFIH1rs7590692 polymorphism（CVC=9/10,test precision=0.53,OR=1.32（95%CI,1.09-1.59）,P<0.01）.Conclusion:In this study,we screened the interaction between susceptibility genes affecting psoriasis and ERAP1 gene and found that the most significant interaction was between ERAP1 rs27044 polymorphism and IFIH1 rs7590692 polymorphism.Psoriasis is a multifactorial autoimmune disease with multiple genetic factors.Although this paper increases understanding of the genetic mechanisms of psoriasis and identifies interactions of susceptibility genes,there are still many unsolved puzzles about the genetic mechanisms of psoriasis.From basic biological studies,it is possible to better understand how these gene interactions influence psoriasis development,and basic research may provide answers to these questions.