Cox-based MDR (CoxMDR) [37] U U U U U No No No No Yes D, Q, MV D D D D No Yes Yes Yes NoMultivariate GMDR (MVGMDR) [38] Robust MDR (RMDR) [39]Blood stress [38] Bladder cancer [39] Alzheimer’s illness [40] Chronic Fatigue Syndrome [41]Log-linear-based MDR (LM-MDR) [40] Odds-ratio-based MDR (OR-MDR) [41] Optimal MDR (Opt-MDR) [42] U NoMDR for Stratified Populations (MDR-SP) [43] UDNoPair-wise MDR (PW-MDR) [44]Simultaneous handling of families and unrelateds Transformation of survival time into dichotomous attribute employing martingale residuals Multivariate modeling applying generalized estimating equations Handling of sparse/empty cells utilizing `unknown risk’ class Enhanced factor mixture by log-linear models and re-classification of threat OR rather of naive Bayes classifier to ?classify its risk Data driven instead of fixed threshold; Pvalues approximated by generalized EVD rather of permutation test Accounting for population stratification by using principal elements; significance estimation by generalized EVD Handling of sparse/empty cells by reducing contingency tables to all doable two-dimensional interactions No D U No DYesKidney transplant [44]NoEvaluation with the classification outcome Extended MDR (EMDR) Evaluation of final model by v2 statistic; [45] consideration of distinctive permutation strategies Various phenotypes or information structures Survival Dimensionality Classification depending on variations beReduction (SDR) [46] tween cell and whole population survival estimates; IBS to evaluate modelsUNoSNoRheumatoid arthritis [46]continuedTable 1. (Continued) Data structure Cov Pheno Modest sample sizesa No No ApplicationsNameDescriptionU U No EAI045 manufacturer QNoSBladder cancer [47] Renal and Vascular EndStage Illness [48] Obesity [49]Survival MDR (Surv-MDR) a0023781 [47] Quantitative MDR (QMDR) [48] U No O NoOrdinal MDR (Ord-MDR) [49] F No DLog-rank test to classify cells; squared log-rank statistic to evaluate models dar.12324 Handling of quantitative phenotypes by comparing cell with overall imply; t-test to evaluate models Handling of phenotypes with >2 classes by assigning every single cell to probably phenotypic class Handling of extended pedigrees working with pedigree disequilibrium test No F No D NoAlzheimer’s disease [50]MDR with Pedigree Disequilibrium Test (MDR-PDT) [50] MDR with Phenomic Analysis (MDRPhenomics) [51]Autism [51]Aggregated MDR (A-MDR) [52]UNoDNoJuvenile idiopathic arthritis [52]Model-based MDR (MBMDR) [53]Handling of trios by comparing quantity of occasions genotype is transmitted GW0918 site versus not transmitted to impacted child; analysis of variance model to assesses effect of Computer Defining significant models using threshold maximizing region below ROC curve; aggregated danger score according to all important models Test of each cell versus all other folks working with association test statistic; association test statistic comparing pooled highrisk and pooled low-risk cells to evaluate models U NoD, Q, SNoBladder cancer [53, 54], Crohn’s disease [55, 56], blood pressure [57]Cov ?Covariate adjustment doable, Pheno ?Feasible phenotypes with D ?Dichotomous, Q ?Quantitative, S ?Survival, MV ?Multivariate, O ?Ordinal.Data structures: F ?Loved ones primarily based, U ?Unrelated samples.A roadmap to multifactor dimensionality reduction methodsaBasically, MDR-based techniques are created for little sample sizes, but some methods give specific approaches to take care of sparse or empty cells, normally arising when analyzing very tiny sample sizes.||Gola et al.Table 2. Implementations of MDR-based techniques Metho.Cox-based MDR (CoxMDR) [37] U U U U U No No No No Yes D, Q, MV D D D D No Yes Yes Yes NoMultivariate GMDR (MVGMDR) [38] Robust MDR (RMDR) [39]Blood stress [38] Bladder cancer [39] Alzheimer’s disease [40] Chronic Fatigue Syndrome [41]Log-linear-based MDR (LM-MDR) [40] Odds-ratio-based MDR (OR-MDR) [41] Optimal MDR (Opt-MDR) [42] U NoMDR for Stratified Populations (MDR-SP) [43] UDNoPair-wise MDR (PW-MDR) [44]Simultaneous handling of families and unrelateds Transformation of survival time into dichotomous attribute employing martingale residuals Multivariate modeling using generalized estimating equations Handling of sparse/empty cells making use of `unknown risk’ class Improved element mixture by log-linear models and re-classification of threat OR as an alternative of naive Bayes classifier to ?classify its risk Data driven rather of fixed threshold; Pvalues approximated by generalized EVD alternatively of permutation test Accounting for population stratification by utilizing principal elements; significance estimation by generalized EVD Handling of sparse/empty cells by reducing contingency tables to all attainable two-dimensional interactions No D U No DYesKidney transplant [44]NoEvaluation with the classification outcome Extended MDR (EMDR) Evaluation of final model by v2 statistic; [45] consideration of distinctive permutation techniques Distinctive phenotypes or data structures Survival Dimensionality Classification according to variations beReduction (SDR) [46] tween cell and whole population survival estimates; IBS to evaluate modelsUNoSNoRheumatoid arthritis [46]continuedTable 1. (Continued) Data structure Cov Pheno Smaller sample sizesa No No ApplicationsNameDescriptionU U No QNoSBladder cancer [47] Renal and Vascular EndStage Disease [48] Obesity [49]Survival MDR (Surv-MDR) a0023781 [47] Quantitative MDR (QMDR) [48] U No O NoOrdinal MDR (Ord-MDR) [49] F No DLog-rank test to classify cells; squared log-rank statistic to evaluate models dar.12324 Handling of quantitative phenotypes by comparing cell with all round mean; t-test to evaluate models Handling of phenotypes with >2 classes by assigning each and every cell to most likely phenotypic class Handling of extended pedigrees employing pedigree disequilibrium test No F No D NoAlzheimer’s illness [50]MDR with Pedigree Disequilibrium Test (MDR-PDT) [50] MDR with Phenomic Evaluation (MDRPhenomics) [51]Autism [51]Aggregated MDR (A-MDR) [52]UNoDNoJuvenile idiopathic arthritis [52]Model-based MDR (MBMDR) [53]Handling of trios by comparing number of instances genotype is transmitted versus not transmitted to affected youngster; analysis of variance model to assesses effect of Pc Defining substantial models employing threshold maximizing location under ROC curve; aggregated risk score according to all considerable models Test of each and every cell versus all others utilizing association test statistic; association test statistic comparing pooled highrisk and pooled low-risk cells to evaluate models U NoD, Q, SNoBladder cancer [53, 54], Crohn’s illness [55, 56], blood stress [57]Cov ?Covariate adjustment achievable, Pheno ?Attainable phenotypes with D ?Dichotomous, Q ?Quantitative, S ?Survival, MV ?Multivariate, O ?Ordinal.Data structures: F ?Household primarily based, U ?Unrelated samples.A roadmap to multifactor dimensionality reduction methodsaBasically, MDR-based solutions are made for compact sample sizes, but some approaches present specific approaches to deal with sparse or empty cells, commonly arising when analyzing incredibly small sample sizes.||Gola et al.Table two. Implementations of MDR-based solutions Metho.