MCPerm (version 1.1.4)

meta.MCPerm: Meta analysis corrected by permutation test

Description

Meta analysis corrected by permutation test.

Usage

meta.MCPerm(case_11, case_12, case_22, control_11, control_12, control_22,  model = "allele", fixed_method = "MH", random_method = "DL", 
	 Qp_alpha = 0.01, repeatNum = 1000)

Arguments

case_11

non-negative integer vector, the frequency of genotype in case samples.

case_12

non-negative integer vector, the frequency of genotype in case samples.

case_22

non-negative integer vector, the frequency of genotype in case samples.

control_11

non-negative integer vector, the frequency of genotype in control samples.

control_12

non-negative integer vector, the frequency of genotype in control samples.

control_22

non-negative integer vector, the frequency of genotype in control samples.

model

a character string indicating the type of model("allele"(default),"dominant" or "recessive") supplied to the data. The risk allele(see details) is marked as allele1. The allele model indicates allele1 versus allele2, the dominant model indicates + versus , the recessive model indicates versus + .

fixed_method

a character string indicating the method('Inverse','MH'(default) or 'Peto') to fit fixed-effects model.

random_method

a character string indicating the method to fit random-effects model: "HE", "DL"(default), "SJ", "ML", "REML", or "EB".

Qp_alpha

the threshold value(0~1)(default 0.01) to refuse the null hypothesis that all studies were evaluating the same effect. Not rejecting the above hypothesis usually leads a meta-analysis to adopt a fixed-effects model. The fixed-effects model assumes that the estimated effect sizes only differ by the sampling error. In contrast, if a significant Q-statistic P

repeatNum

an integer(default 1000) specifying the number of replicates used in the Monte Carlo permutation.

Value

corrected_result

matrix with 3 rows and 3 columns. The first column is Cochran's Q-statistics value; the second column is I2(=max(Q-df/Q,0), the degree of inconsistency across studies); the third column is the merged log odds ratio. The first row is the value for the 3 columns calculated by true data of the studies; the second row is the p value calculated by true data of the studies; the third row is the p value calculated by permutation data, namely equal to the proportion of the statistic of interest as extreme as the observed value of the same statistic. Note: the p value for I2 donot calculate throngh meta-analysis, so the value is NA.

risk_allele

the symbol of risk allele.See details.

true_merged_LnOR

the merged log odd ratio by meta analysis using observation data.

true_merged_LnOR_VAR

variation of the merged log odd ratio by meta analysis using observation data.

true_merged_LnOR_p

p value for the merged log odd ratio by meta analysis using observation data.

true_merged_LnOR_ci.lb

lower bound of the confidence intervals for the merged log odds ratio.

true_merged_LnOR_ci.ub

upper bound of the confidence intervals for the merged log odds ratio.

study_num

the number of studies in the meta analysis.

sample

a vector with 'study_num' elements, the sample size of each study.

true_LnOR

a vector of log odd ratio calculated by each study.

true_VARLnOR

a vector, the varition of log odd ratio calculated by each study.

perm_case_11,perm_case_12,perm_case_22,perm_control_11,perm_control_12,perm_control_22

a matrix with one row and 'repeatNum' columns, the freqency for genotype got by permutating data by MCPerm method. MCPerm method details see chisq.MCPerm.

perm_LnOR

matirx with rows indicating studies and 'repeatNum' columns indicating log odds ratio for each permutation data.

perm_VARLnOR

matrix with rows indicating studies and 'repeatNum' columns indicating variance for the log odds ratio for each permuation data.

perm_Qp

a matrix with one row and 'repeatNum' columns, statistic Q value of heterogeneity for each permutation data.

perm_I2

a matrix with one row and 'repeatNum' columns, statistic I2 value of heterogeneity for each permutation data.

perm_merged_LnOR

a matrix with one row and 'repeatNum' columns, merged log odd ratio for permutation data.

perm_merged_VARLnOR

a matrix with one row and 'repeatNum' columns, variation of merged log odd ratio for permuation data.

perm_p

a matrix with one row and 'repeatNum' columns, p value for merged log odd ratio of permuation data.

model,fixed_method,random_method,Qp_alpha,repeatNum

value for parmeter of the funciton.

Details

Allele 1 and allele 2 to each study have OR values. The risk allele is the allele which the number of studies which OR>1 more than half of the number of all studies.

MCPerm details see chisq.MCPerm.

References

William S Noble(Nat Biotechnol.2009): How does mutiple testing correction work?

Hedges,L.V. & Vevea,J.L.(1998): Fixed- and random- effects models in meta-analysis.

See Also

meta, meta.TradPerm, permuteGenotype, permuteGenotypeCount, genotypeStat, OR.TradPerm, Armitage.TradPerm, chisq.TradPerm, fisher.TradPerm, VS.Genotype.Hist, VS.Allele.Hist, VS.Hist, PermMeta.LnOR.Hist, PermMeta.LnOR.CDC, PermMeta.LnOR.boxplot, PermMeta.boxplot, PermMeta.Hist, pearson_scatter, Q.TradPerm, I2.TradPerm

Examples

## import data
# data(MetaGenotypeCount)
## delete the first line which is the names for columns.
# temp=MetaGenotypeCount[-1,,drop=FALSE]
# result=meta.MCPerm(case_11=as.numeric(temp[,14]),case_12=as.numeric(temp[,16]),
	 # case_22=as.numeric(temp[,18]),control_11=as.numeric(temp[,15]),
	 # control_12=as.numeric(temp[,17]),control_22=as.numeric(temp[,19]),
	 # model="allele",method="MH",repeatNum=100000)
# result