Limits...
Gene Network Reconstruction by Integration of Prior Biological Knowledge.

Li Y, Jackson SA - G3 (Bethesda) (2015)

Bottom Line: The a priori information can be calculated or retrieved from other biological data, e.g., Gene Ontology similarity, protein-protein interaction, gene regulatory network.By incorporating prior knowledge, the weighted graphical lasso (wglasso) outperforms the original glasso both on simulations and on data from Arabidopsis.Simulation studies show that even when some prior knowledge is not correct, the overall quality of the wglasso network was still greater than when not incorporating that information, e.g., glasso.

View Article: PubMed Central - PubMed

Affiliation: Center for Applied Genetic Technologies, University of Georgia, Athens, Georgia Institute of Plant Breeding, Genetics and Genomics, University of Georgia, Athens, Georgia Department of Statistics, University of Georgia, Athens, Georgia 30602.

No MeSH data available.


Simulation to compare glasso and wglasso. The simulation was repeated 100 times for each combination of sample sizes (n) and precision ratio of the prior information. Gene number = 100. In each simulation, the maximum Matthews correlation coefficient (maxMCC) of estimated networks from different penalty parameters is recorded. The Y-axis shows the mean of maxMCC from the 100 simulations. The error bars are 95% confidence intervals.
© Copyright Policy - open-access
Related In: Results  -  Collection

License
getmorefigures.php?uid=PMC4478538&req=5

fig3: Simulation to compare glasso and wglasso. The simulation was repeated 100 times for each combination of sample sizes (n) and precision ratio of the prior information. Gene number = 100. In each simulation, the maximum Matthews correlation coefficient (maxMCC) of estimated networks from different penalty parameters is recorded. The Y-axis shows the mean of maxMCC from the 100 simulations. The error bars are 95% confidence intervals.

Mentions: Additional simulations were performed with a variety of sample sizes and precision ratios in order to systematically evaluate wglasso. The simulation was repeated 100 times for each combination of sample size and precision ratio. The results show that reconstructed networks with a priori information had a significantly greater maxMCC than without a priori information, indicating superior performance of wglasso (Figure 3). In most cases, wglasso outperformed glasso, even when most of the prior knowledge was incorrect. If the sample size was high and the precision ratio low, e.g., sample size = 300 and precision ratio = 0.2, it is possible that an excess of incorrect prior information would be harmful. However, in real situations, the sample size is often much lower than the gene number and misleading a priori information based on experimental studies is likely to be low. Moreover, highly accurate prior knowledge results in more accurately reconstructed networks. In practice, it would be possible to integrate multiple resources to increase the reliability of prior information.


Gene Network Reconstruction by Integration of Prior Biological Knowledge.

Li Y, Jackson SA - G3 (Bethesda) (2015)

Simulation to compare glasso and wglasso. The simulation was repeated 100 times for each combination of sample sizes (n) and precision ratio of the prior information. Gene number = 100. In each simulation, the maximum Matthews correlation coefficient (maxMCC) of estimated networks from different penalty parameters is recorded. The Y-axis shows the mean of maxMCC from the 100 simulations. The error bars are 95% confidence intervals.
© Copyright Policy - open-access
Related In: Results  -  Collection

License
Show All Figures
getmorefigures.php?uid=PMC4478538&req=5

fig3: Simulation to compare glasso and wglasso. The simulation was repeated 100 times for each combination of sample sizes (n) and precision ratio of the prior information. Gene number = 100. In each simulation, the maximum Matthews correlation coefficient (maxMCC) of estimated networks from different penalty parameters is recorded. The Y-axis shows the mean of maxMCC from the 100 simulations. The error bars are 95% confidence intervals.
Mentions: Additional simulations were performed with a variety of sample sizes and precision ratios in order to systematically evaluate wglasso. The simulation was repeated 100 times for each combination of sample size and precision ratio. The results show that reconstructed networks with a priori information had a significantly greater maxMCC than without a priori information, indicating superior performance of wglasso (Figure 3). In most cases, wglasso outperformed glasso, even when most of the prior knowledge was incorrect. If the sample size was high and the precision ratio low, e.g., sample size = 300 and precision ratio = 0.2, it is possible that an excess of incorrect prior information would be harmful. However, in real situations, the sample size is often much lower than the gene number and misleading a priori information based on experimental studies is likely to be low. Moreover, highly accurate prior knowledge results in more accurately reconstructed networks. In practice, it would be possible to integrate multiple resources to increase the reliability of prior information.

Bottom Line: The a priori information can be calculated or retrieved from other biological data, e.g., Gene Ontology similarity, protein-protein interaction, gene regulatory network.By incorporating prior knowledge, the weighted graphical lasso (wglasso) outperforms the original glasso both on simulations and on data from Arabidopsis.Simulation studies show that even when some prior knowledge is not correct, the overall quality of the wglasso network was still greater than when not incorporating that information, e.g., glasso.

View Article: PubMed Central - PubMed

Affiliation: Center for Applied Genetic Technologies, University of Georgia, Athens, Georgia Institute of Plant Breeding, Genetics and Genomics, University of Georgia, Athens, Georgia Department of Statistics, University of Georgia, Athens, Georgia 30602.

No MeSH data available.