SynTReN: A Tool for Generating Synthetic Gene Expression Data
Author Information
Author(s): Van den Bulcke Tim, Van Leemput Koenraad, Naudts Bart, Remortel Piet, Ma Hongwu, Verschoren Alain, De Moor Bart, Marchal Kathleen
Hypothesis
The study proposes a network generator that creates synthetic transcriptional regulatory networks to benchmark structure learning algorithms.
Conclusion
The network generation technique offers a valid alternative to existing methods, producing synthetic gene expression data that closely resembles real biological networks.
Supporting Evidence
- The generator produces synthetic data that approximates experimental data.
- The statistical properties of the generated networks closely resemble those of genuine biological networks.
- The tool allows for user-definable parameters to adjust the complexity of the resulting data set.
Takeaway
This study created a tool that makes fake gene data that looks like real data, helping scientists test their algorithms better.
Methodology
The tool generates synthetic transcriptional regulatory networks and corresponding microarray data by selecting subnetworks from known biological networks and modeling interactions using Michaelis-Menten and Hill kinetics.
Potential Biases
The generated networks may be biased towards well-studied pathways due to the reliance on previously characterized networks.
Limitations
The method relies on the accuracy of the source networks and may not capture all subtle characteristics of real networks.
Digital Object Identifier (DOI)
Want to read the original?
Access the complete publication on the publisher's website