Editorial Statistical Genetics and Its Applications in Medical ...downloads.hindawi.com/journals/cmmm/2014/712073.pdfEditorial Statistical Genetics and Its Applications in Medical

EditorialStatistical Genetics and Its Applications in Medical Studies

Ao Yuan,1 Wenqing He,2 Gengsheng Qin,3 and Qizhai Li4

1 Department of Biostatistics, Bioinformatics and Biomathematics, Georgetown University, Washington, DC 20057, USA2Department of Statistics and Actuarial Science, University of Western ontario, London, ON, Canada N6A 5B73Department of Mathematics and Statistics, Georgia State University, Atlanta, GA 30303, USA4Academy of Mathematics and Systems Science, Chinese Academy of Sciences, Beijing 100190, China

Correspondence should be addressed to Ao Yuan; [email protected]

Received 8 December 2013; Accepted 9 December 2013; Published 6 March 2014

Copyright © 2014 Ao Yuan et al.This is an open access article distributed under the Creative Commons Attribution License, whichpermits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

Statistical genetics can be viewed as a classical branch ofapplied probability and statistics, which has recently gainedmuch momentum, due to the significant breakthroughs ingenetics. With the availability of modern techniques, newmethods, and significantly increased data information, it isimperative to study the relationship between gene traits suchas diseases and genetic susceptibilities in an unprecedentedmanner. This area is among the hottest topics in appliedstatistics, applied mathematics, biological/medical studies,and other related sciences.

The main aim of this special issue focuses on thenew development and applications of computational, math-ematical, and statistical methods in genetic disease study.The special issue could become an international forum forresearchers to exchange their new thoughts, most recentdevelopments, and ideas in the field.

In this special issue, we selected seven articles within theabove topics. Below is a brief summary of the selected articlesin this special issue.

Recent advances in biotechnologies have led to theidentification of an enormous number of genetic markersin disease association studies; how to select a smaller setof genes to explore the relation between genes and diseaseis a challenging task. Bayesian methods have the advantageof incorporating prior information into the model for suchanalysis. Article “Applications of bayesian gene selection andclassification with mixtures of generalized singular g-priors” byW.-K. Chien and C. K. Hsiao addresses this problem usingBayesian method with a Gaussian prior and inverse gammahyperprior. The proposed approach is applied to a colonand leukemia cancer study. Comparison with other existing

methods was conducted. The authors find that classificationaccuracy of the proposedmodel is higher with a smaller set ofselected genes and that the results not only replicated findingsin several earlier studies, but also provided the strength ofassociation with posterior probabilities.

Article “Modified logistic regression models using genecoexpression and clinical features to predict prostate cancerprogression” by H. Zhao et al. proposed a new logisticregression model for predicting prostate cancer progression.They incorporated coexpressed gene profiles into the logisticmodel based on clinical features to improve the inferenceaccuracy. Then they use the top-scoring pair method toselect genes with significant association with the disease. Theperformance of the proposedmethod is compared with somecommonly used methods for such problem, using data setsfrom such published studies. Their study suggests that theproposed method performs better than a commonly usedone and that the top-scoring pair method is a useful toolfor feature (and/or gene) selection to be used in prognosticmodels.

Resampling-based multiple testing procedures are widelyused in genomic studies to identify differentially expressedgenes and for genome-wide association studies. The powerand stability of these popular procedures have not beenextensively evaluated. Article “Power and stability propertiesof resampling-based multiple testing procedures with appli-cations to gene oncology studies” by D. Li and T. D. Dyeinvestigates the power and stability of seven commonlyused resampling-based multiple testing procedures that arefrequently used in high-throughput data analysis for smallsample size data. Simulations and real data gene oncology

Hindawi Publishing CorporationComputational and Mathematical Methods in MedicineVolume 2014, Article ID 712073, 3 pageshttp://dx.doi.org/10.1155/2014/712073

2 Computational and Mathematical Methods in Medicine

examples are employed in their investigation. Their studysuggests that the bootstrap single-step minP procedure andthe bootstrap step-down minP procedure perform the best,when sample size is as small as 3 in each group and eitherfamilywise error rate or false discovery rate control is desired.When sample size increases to 12 and false discovery ratecontrol is desired, the permutation maxT procedure and thepermutation minP procedure perform the best.

Article “Transcriptional protein-protein cooperativity inPOU/HMG/DNA complexes revealed by normal mode anal-ysis” by D. D. Wang and H. Yan investigates how proteins inPOU/HMG/DNA ternary complexes interact cooperatively,which are crucial in transcriptional regulation of embryonicstem cells. They use the normal mode analysis to detectthe most cooperative or collective motions (essential modes)of a large number of proteins, a commonly used tool toanalyze the structural dynamics of biomolecules, whichcombines some techniques in engineering, mathematics, andstatistics. Their work reveals how the two proteins Oct-1and Sox-2 work together physically and structurally at twospecificDNAbiding sites, by analyzing themotionmagnitudefunctions. A correlation measure is used to characterize theamount of cooperativity of pairs of proteins. The proposedmethods provide useful information for understanding thecomplicated interactionmechanism in the POU/HMG/DNAcomplexes. The corresponding online computational toolsare also provided.

In modern medical diagnosis or genetic studies, thereceiver operating characteristic (ROC) curve is a populartool to evaluate the discrimination performance of biomark-ers on a disease status or a phenotype. With the presenceof a number of covariates in the data, how to select themost relevant covariables, or to select the model with goodoverall properties, is a challenging problem. Article “Variableselection in ROC regression” by B. Wang addresses thisproblem with an interesting idea. There are a large numberof criteria available for this problem.The author first rewritesthe ROC regression into a grouped variable selection formso that current criteria can be applied and then proposesa general two-stage framework with a BIC selector for thegroup SCAD algorithm under the local model assumption.Basic asymptotic properties of the proposed methods arederived. Simulation studies and real data analysis showthat the proposed grouped variable selection is superior tothe traditional model selections. Furthermore, the authorfinds that the focused information criterion provides moreaccurate estimated area under the ROC curve compared withother criteria.

Two-stage design and analysis are often adopted ingenome-wide association studies (GWASs). Consideringthe genetic model uncertainty, many robust procedureshave been proposed and applied in GWASs. The existingapproaches mostly focused on binary traits, and many ofthese methods analyze data based on two separate stages, andfew work has been done on continuous (quantitative) traits.Article “Robust joint analysis with data fusion in two-stagequantitative trait genome-wide association studies” by D.-D.Pan et al. proposes a powerful F-statistic-based robust jointanalysis method for quantitative traits using the combined

raw data from both stages, in which the genetic effects aremodeled as regression parameters. Variations of the MAXtesting statistic are constructed to calculate the statisticalsignificance and power. It is well known that critical valuesand power of the MAX type statistic are not easy to compute.The authors derived analytic expressions on the basis of theasymptotic distributions, so that these quantities can be easilyobtained. They show using simulations that the proposedmethod is substantially more robust than the 𝐹-test basedon the commonly used additive model when the underlyinggenetic model is unknown.

Multiple meta-analyses may use similar search criteriaand focus on the same topic of interest, but they mayyield different or sometimes discordant results. The lack ofstatistical methods for synthesizing these findings makes itchallenging to properly interpret the results from multiplemeta-analyses, especially when their results are conflicting.Article “A statistical method for synthesizing meta-analyses”by L. L. Tang et al. introduces a method to synthesize themeta-analytic results under two cases: (1) when multiplemeta-analyses use the same type of summary effect estimatesand (2) when meta-analyses use different types of effectsizes. In case 2, the meta-analysis results cannot be directlycombined; therefore they propose a two-step frequentistprocedure to first convert the effect size estimates to thesamemetric and then summarize themwith aweightedmeanestimate.The proposed method has the following advantagesover some existing methods: different types of summaryeffect sizes can be considered; the same overall effect size canbe provided by conducting a meta-analysis on all individualstudies from multiple meta-analyses.

One of the main objectives of a genome-wide associationstudy (GWAS) is to develop a prediction model for a binaryclinical outcome using single-nucleotide polymorphisms(SNPs) which can be used for diagnostic and prognosticpurposes and for better understanding of the relationshipbetween the disease and SNPs. Penalized support vectormachine (SVM) methods have been widely used toward thisend. However, since investigators often ignore the geneticmodels of SNPs, a final model results in a loss of efficiency inprediction of the clinical outcome. Article “SNP selection ingenome-wide association studies via penalized support vectormachine with MAX test” by J. Kim et al. proposes a two-stage method such that the genetic models of each SNP areidentified using the MAX test and then a prediction modelis fitted using a penalized SVM method. They apply the pro-posed method to various penalized SVMs and compare theirperformances using various penalty functions. They show bysimulations and real GWAS data analysis that the proposedmethod performs better than the prediction methods thatignore the genetic models, in terms of prediction power andselectivity.

Using DNA sequence data in the study of ancestralhistory of human population is an essential part in theunderstanding of human evolution. The existing methodsfor such coalescence inference using the method of eitherthe rooted tree or unrooted tree constructed from theobserved data, both of which use recursion formulae tocompute the data probabilities. These methods are useful in

Computational and Mathematical Methods in Medicine 3

practical applications but computationally complicated. Arti-cle “On coalescence analysis using genealogy rooted trees” byA. Yuan et al. explores a new method for this problem. Theyfirst investigate the asymptotic behavior of such inference;their results indicate that, broadly, the estimated coalescenttime will be consistent to a finite limit. Then they study arelatively simple computation method for this analysis andillustrate how to use it.

Acknowledgment

We thank all the authors who contributed to this special issue.

Ao YuanWenqing He

Gengsheng QinQizhai Li

Submit your manuscripts athttp://www.hindawi.com

Stem CellsInternational

Hindawi Publishing Corporationhttp://www.hindawi.com Volume 2014


MEDIATORSINFLAMMATION

of


Behavioural Neurology

EndocrinologyInternational Journal of



Disease Markers


BioMed Research International

OncologyJournal of



Oxidative Medicine and Cellular Longevity


PPAR Research

The Scientific World JournalHindawi Publishing Corporation http://www.hindawi.com Volume 2014

Immunology ResearchHindawi Publishing Corporationhttp://www.hindawi.com Volume 2014

Journal of

ObesityJournal of



Computational and Mathematical Methods in Medicine

OphthalmologyJournal of


Diabetes ResearchJournal of



Research and TreatmentAIDS


Gastroenterology Research and Practice


Parkinson’s Disease

Evidence-Based Complementary and Alternative Medicine

Volume 2014Hindawi Publishing Corporationhttp://www.hindawi.com

Editorial Statistical Genetics and Its Applications in Medical ...downloads.hindawi.com/journals/cmmm/2014/712073.pdfEditorial Statistical Genetics and Its Applications in Medical

Documents