Exploring the feasibility of next-generation sequencing and microarray data meta-analysis

Abstract

Emerging next-generation sequencing (NGS) technology potentially resolves many issues that prevent widespread clinical use of gene expression microarrays. However, the number of publicly available NGS datasets is still smaller than that of microarrays. This paper explores the possibilities for combining information from both microarray and NGS gene expression datasets for the discovery of differentially expressed genes (DEGs). We evaluate several existing methods in detecting DEGs using individual datasets as well as combined NGS and microarray datasets. Results indicate that analysis of combined NGS and microarray data is feasible, but successful detection of DEGs may depend on careful selection of algorithms as well as on data normalization and pre-processing.

Fig. 1

Assessment of microarray and NGS meta-analysis methods involves (1) pre-processing of raw data to calculate gene expression and (2) application of several platform-specific DEG detection methods.

Fig. 2

False discovery rate versus the number of DEGs identified in NGS data using various NGS-specific methods.

Fig. 3

False discovery rate versus the number of DEGs identified in microarray data using various microarray-specific methods.

Fig. 4

False discovery rate versus the number of DEGs identified using Rank Products applied to individual NGS and microarray data as well as to combined data.