Home|Sandbox Demo| Research Projects|Tutorial|Documentation| FAQs

Project: miniTUBA Demo No. 2

Investigators: miniTUBA team at University of Michigan. Created at: 2007-01-24 16:31. Last updated at: 2007-02-14 15:55.


This Demo project explores how miniTUBA can identify causal relationships among biomedical variables. We assume four proteins (e.g., cytokines) in time course biomedical setting have known causal relationships among them. Based on the giving four protein (A-D) network and their probability table, a synthetic dataset is dynamically generated by a computer script. The initial states (time point 0) are randomly picked up, and the values of different time points are sampled from randomly pre-assigned conditional probability distribution. The structure prior knowledge is set as protein A can not have any parents.

This is a miniTUBA Demo project. Users can view this project but cannot change the settings and analysis results. For this demo, the following is the four protein network, probability table, and synthetic data based on the probability table. Our miniTUBA will show how to use the synthetic data to identify the same protein network.

Network Topology: Probability Tables:
P(0.5<Protein1≤1) P(0≤Protein1<0.5)
0.5 0.5
Protein1 P(0.5<Protein4≤1|Protein1) P(0≤Protein4<0.5|Protein1)
0.5<Protein1≤1 0.6 0.4
0≤Protein1<0.5 0.1 0.9
Protein2 P(0.5<Protein3≤1|Protein2) P(0≤Protein3<0.5|Protein2)
0.5<Protein2≤1 0.8 0.2
0≤Protein2<0.5 0.9 0.1

Synthetic Data:

Click here to download the data file.
Protein3 P(0.5<Protein2≤1|Protein3) P(0≤Protein2<0.5|Protein3)
0.5<Protein3≤1 0.3 0.7
0≤Protein3<0.5 0.1 0.9

Status: Approved


View Data The detailed analysis operations are only accessible after Login with a free demo account(user name: demo@e.d.u, password: demo).

Analyses: Refresh

No. ID Status Note Show Settings Show Results Remove Results Allow Others to View
2 277 Finished Test No. 2 Show settings Show results Remove results
1 276 Finished Test No. 1 Show settings Show results Remove results
Login|Register| Acknowledgements| Links|Contact Us|Disclaimer