High-dimensional genomic data bias correction and data integration using MANCIE.

Chongzhi Zang,X Shirley Liu,Tao Wang,Housheng Hansen He,Shihua Zhang,Sheng’En Hu,Qian Qin,Tengfei Xiao,Myles Brown,Ke Deng,Clifford A Meyer,Jun S Liu,Yang Xie,Bo Li

doi:10.1038/ncomms11305

Abstract

High-dimensional genomic data analysis is challenging due to noises and biases in high-throughput experiments. We present a computational method matrix analysis and normalization by concordant information enhancement (MANCIE) for bias correction and data integration of distinct genomic profiles on the same samples. MANCIE uses a Bayesian-supported principal component analysis-based approach to adjust the data so as to achieve better consistency between sample-wise distances in the different profiles. MANCIE can improve tissue-specific clustering in ENCODE data, prognostic prediction in Molecular Taxonomy of Breast Cancer International Consortium and The Cancer Genome Atlas data, copy number and expression agreement in Cancer Cell Line Encyclopedia data, and has broad applications in cross-platform, high-dimensional data integration.

Highlights

High-dimensional genomic data analysis is challenging due to noises and biases in high-throughput experiments
The two data matrices contain profiles on the same set of samples generated using different experimental platforms (for example, copy number variation (CNV) and RNA-seq on the same collection of tumours), or generated independently
If the rows of the two matrices are unmatched, MANCIE first generates a summarized associated matrix that has matched rows with the main matrix using a biologically motivated matching process (Supplementary Fig. 1, see Methods for details). This matching step requires additional biological information to connect the rows between the two matrices, for example, each gene will corresponds to a row vector summarized from a few nearby transcription factor (TF) -binding sites

Summary

Introduction

High-dimensional genomic data analysis is challenging due to noises and biases in high-throughput experiments. We present a computational method matrix analysis and normalization by concordant information enhancement (MANCIE) for bias correction and data integration of distinct genomic profiles on the same samples. By computational analyses of these high-dimensional data matrices using dimension reduction (for example, principal component analysis, PCA) or clustering approaches, one can learn characteristic information within samples and identify key features between samples to interrogate biological functions. Surrogate variable analysis (SVA)[9] models the gene-expression heterogeneity bias as ‘surrogate variables’ and separate them from primary variables that capture biologically meaningful information These methods aim to normalize data within the same data matrix from the same platform. Applied to ENCODE, METABRIC, TCGA and CCLE data, MANCIE showed effectiveness in improved identification of biologically meaningful patterns

Methods

Results

Conclusion