calculateESpath: Calculation of Effects Sizes and their variance for the...

In GSEMA: Gene Set Enrichment Meta-Analysis

Calculation of Effects Sizes and their variance for the different Gene Sets and studies

Description

This function uses different estimators to calculate the different effects size and their variance for each gene and for each dataset

Usage

calculateESpath(
  objectMApath,
  measure = c("limma", "SMD", "MD"),
  WithinVarCorrect = TRUE,
  missAllow = 0.3
)

Arguments

objectMApath	A list of list. Each list contains two elements. The first element is the Gene Set matrix (gene sets in rows and samples in columns) and the second element is a vector of zeros and ones that represents the state of the different samples of the Gene Sets matrix. 0 represents one group (controls) and 1 represents the other group (cases).
measure	A character string that indicates the type of effect size to be calculated. The options are "limma", "SMD" and "MD". The default value is "limma". See details for more information.
WithinVarCorrect	A logical value that indicates if the within variance correction should be applied. The default value is TRUE. See details for more information.
missAllow	a number that indicates the maximum proportion of missing values allowed in a sample. If the sample has more proportion of missing values the sample will be eliminated. In the other case the missing values will be imputed using the K-NN algorithm.

Details

The different estimator methods that can be applied are:

1. "limma"

2. "SMD"

3. "MD"

The "SMD" (Standardized mean different) method calculates the effect size using the Hedges'g estimator (Hedges, 1981).

The "MD" (raw mean different) calculates the effects size as the difference between the means of the two groups (Borenstein, 2009).

The "limma" method used the limma package to calculate the effect size and the variance of the effect size. The effect size is calculated from the moderated Student's t computed by limma. From it, the estimator of Hedges'g and its corresponding variance are obtained based on (Rosenthal, R., & Rosnow, R. L., 2008)) In this way, some of the false positives obtained by the "SMD" method are reduced.

The WithinVarCorrect parameter is a logical value that indicates if the within variance correction should be applied. In the case of applying the correction, the variance of the gene sets in each of the studies is calculated based on the mean of the estimators and not on the estimator of the study itself as described in formula (21) by (Lin L and Aloe AM 2021.)

Value

A list formed by two elements:

• First element (ES) is a dataframe were columns are each of the studies (datasets) and rows are the genes sets. Each element of the dataframe represents the Effect Size.

• Second element (Var) is a dataframe were columns are each of the studies (datasets) and rows are the genes sets. Each element of the dataframe represents the variance of the Effect size.

Author(s)

Juan Antonio Villatoro Garcia, [e-mail]

References

Borenstein, M. (2009). Effect sizes for continuous data. In H. Cooper, L. V. Hedges, & J. C. Valentine (Eds.), The handbook of research synthesis and meta-analysis (2nd ed., pp. 221–235). New York: Russell Sage Foundation.

Hedges, L. V. (1981). Distribution theory for Glass's estimator of effect size and related estimators. Journal of Educational Statistics, 6(2), 107–128. \Sexpr[results=rd]{tools:::Rd_expr_doi("doi:10.2307/1164588")}

Lin L, Aloe AM (2021). Evaluation of various estimators for standardized mean difference in meta-analysis. Stat Med. 2021 Jan 30;40(2):403-426. \Sexpr[results=rd]{tools:::Rd_expr_doi("10.1002/sim.8781")}

Rosenthal, R., & Rosnow, R. L. (2008). Essentials of behavioral research: Methods and data analysis. McGraw-Hill.

See Also

createObjectMApath

Examples

data("simulatedData")
calculateESpath(objectMApath = objectMApathSim, measure = "limma")

GSEMA documentation built on Oct. 14, 2024, 5:09 p.m.