Authors:
Kathryn M. Dempsey
and
Hesham H. Ali
Affiliation:
University of Nebraska at Omaha, United States
Keyword(s):
Correlation Networks, Network Stability.
Related
Ontology
Subjects/Areas/Topics:
Bioinformatics
;
Biomedical Engineering
;
Computational Molecular Systems
;
Model Design and Evaluation
;
Systems Biology
Abstract:
Recent progress in high-throughput technology has resulted in a significant data overload. Determining how
to obtain valuable knowledge from such massive raw data has become one of the most challenging issues in
biomedical research. As a result, bioinformatics researchers continue to look for advanced data analysis
tools to analysis and mine the available data. Correlation network models obtained from various biological
assays, such as those measuring gene expression levels, are a powerful method for representing correlated
expression. Although correlation does not always imply causation, the correlation network has been shown
to be effective in identifying elements of interest in various bioinformatics applications. While these models
have found success, little to no investigation has been made into the robustness of relationships in the
correlation network with regard to vulnerability of the model according to manipulation of sample values.
Particularly, reservations about the corr
elation network model stem from a lack of testing on the reliability
of the model. In this work, we probe the robustness of the model by manipulating samples to create six
different expression networks and find a slight inverse relationship between sample count and network
size/density. When samples are iteratively removed during model creation, the results suggest that network
edges may or may not remain within the statistical parameters of the model, suggesting that there is room
for improvement in the filtering of these networks. A cursory investigation into a secondary robustness
threshold using these measures confirms the existence of a positive relationship between sample size and
edge robustness. This work represents an important step toward better understanding of the critical noise
versus signal issue in the correlation network model.
(More)