Robust Microarray Meta-Analysis Identifies Differentially Expressed Genes for Clinical Prediction

Table 3

Properties of six microarray meta-analysis methods.

	Rank average	mDEDS	Rank products	Choi	Wang	Naive (control)
Basic FS methods considered	FC, 𝑇, SAM, RS, mRMRQ/D	FC, 𝑇, SAM, RS, mRMRQ/D	FC^1	𝑇^2	FC^3	FC, 𝑇, SAM, RS, mRMRQ/D
Chooses data-specific basic FS method(s)	Yes	No	No	No	No	Yes
Rank-Based	Yes	No	Yes	No	No	No

¹Fold change between all interclass pairs of samples. ²Most similar to a t-statistic, but includes an estimate of interstudy variation. ³Computes a variance-weighted average of fold change. FC: Fold Change, T = t-statistic (t-test), SAM: significance analysis of microarrays, RS: rank sum test, mRMRQ/D: minimum redundancy, maximum relevance with quotient/difference.