8.The functional influence of protein on gene regulatory. Stress
responses were believed to be predominantly regulated at the
transcriptional level; however, the adaptive mechanisms should
include post-transcriptional and post-translational events. To
address this issue, three layers of regulation have been
integrated as transcriptome, translatome, and proteome,
which is useful to gain a deeper understanding of how sophis-
ticated regulation networks operate [99]. And semi-supervised
normalization pipelines have been designed and performed
experimental characterization to create a quality-controlled
multi-omics compendium forE. coli, and a multi-scale model
has further been trained by integrating four omics layers to
predict genome-wide concentrations and growth
dynamics [100].3.3 Top-Down
Integration
The standard “bottom-up integration” approach as above integra-
tive clustering is usually to separate clustering followed by manual
integration. By contrast, a more computational powerful approach
would incorporate all data types simultaneously and generate a
single integrated cluster assignment (see Note 3), which are
thought as “top-down integration” as shown in Table4.1.Statistic-based integration model. One key integrative idea is
unifying hidden factor from different types of data. A joint
latent variable model as iCluster is developed for integrative
clustering by incorporating flexible modeling of the associa-
tions between different data types and the variance-covariance
structure within data types while simultaneously reducing the
dimensionality of the datasets [24]. To extend the scope of
integrative analysis for the inclusion of somatic mutation data,
an expanded framework iCluster+ is further proposed to
ensemble discrete and continuous variables that arise from
integrated genomic, epigenomic, and transcriptomic profiling
[11]. Similarly, a novel algorithm termed moCluster employs a
multiblock multivariate analysis to define a set of latent vari-
ables representing joint patterns across input datasets, which is
passed to an ordinary clustering algorithm in order to discover
joint clusters [101]. The other important integrative idea is
unifying data distribution under the theoretical framework
around Bayesian principles. An integrative Bayesian analysis of
genomics data (iBAG) framework is proposed to identify
important genes/biomarkers by using hierarchical modeling
to combine the data obtained from multiple platforms into
one model [25]. And a Bayesian method referred as MDI
(Multiple Dataset Integration) has been presented for the
unsupervised integrative modeling, where each dataset is mod-
eled using a Dirichlet-multinomial allocation (DMA) mixture
model, with dependencies between these models capturedIntegrative Analysis of Omics Big Data 121