Computational Systems Biology Methods and Protocols.7z

e.g., an element in such tensor can reflect the expression value of

one gene of the same individual at an early or latter time point

during virus infection [69]. Furthermore, nowadays, known as the

era of big data, more delicate biological experiments can be carried

on, and more complex data structure would be faced, e.g., the

combination of tensor as “gene-sample-source-time”, whose rep-

resentative case is the cross-tissue gene expression analysis on

evolutionary [70].

3 Methods

3.1 Batch Effect
Removal Before
Integration

The removal of batch effect should be an important step ahead of

many integrative analyses on biological big data. Many variables will

play in any given research, such as the influence of age or sex on the

diseases. Especially there are many sources of variation when the

expressions of thousands of genes are measured at once, so that the

batch effects become more critical due to the complexity of gen-

omes inside and environments outside [71]. In practice, the

sequencing and microarray samples are usually collected or pro-

cessed in multiple batches (e.g., at different times), which are easy

to produce technical biases and incorrect results in the downstream

analysis [72]. For example, to estimate expression correlation over

Fig. 1The organization of data structures in omics big data

Integrative Analysis of Omics Big Data 115