In the second DREAM4 large subchallenge, DREAM4 size 100 multifactorial, only multifactorial data were available, therefore all the methods were included in the comparison, and run as originally designed.
In this work, we have combined multifactorial data to identify network-based biomarkers.
Wildtype, knockout, knockdown, and multifactorial data describe the expression of initially perturbed genes, which are however in a steady state at the time of measurement, whereas time series data describe the dynamics of the expression levels of initially perturbed genes.
Extensive, multifactorial data sharing is a crucial prerequisite for current and future (radiotherapy) research.
Network 5 demonstrates the utility of the multifactorial data (Fig. 9) for network reconstruction.
Multifactorial data might correspond for example to expression profiles obtained from different patients or biological replicates.
The four-gene cycle (genes 5→6→8→7→5, Fig. 8A) in network 5 is an example for a difficult network motif that our approach predicts correctly only if the multifactorial data is included.
Knockout data are more informative for network inference than knockdown or multifactorial data.
In this work, we have used a network-based strategy to optimize the discovery of biomarkers using multifactorial data, including patient expression, clinical survival, and protein-protein interaction (PPI) data.
An acknowledged caveat to the inclusion of pathology data in multifactorial likelihood modeling is the underlying assumption that missense and in-frame deletions considered to be pathogenic mutations will exhibit the same tumor histopathological characteristics as do truncating mutations.
