Validation of gene regulatory network inference based on controllability.
Bottom Line:
The reasoning behind such a criterion is that, if our purpose is to use gene regulatory networks to design therapeutic intervention strategies, then we are not concerned with network fidelity, per se, but only with our ability to design effective interventions based on the inferred network.The objective of a control policy is to optimally reduce the total steady-state probability mass of the undesirable states (phenotypes), which is equivalent to optimally increasing the total steady-state mass of the desirable states.Hence, when one is aiming at a specific application, it may be wise to use an objective-based measure of inference validity.
View Article:
PubMed Central - PubMed
Affiliation: Department of Electrical and Computer Engineering, Texas A & M University TX, USA.
ABSTRACT
There are two distinct issues regarding network validation: (1) Does an inferred network provide good predictions relative to experimental data? (2) Does a network inference algorithm applied within a certain network model framework yield networks that are accurate relative to some criterion of goodness? The first issue concerns scientific validation and the second concerns algorithm validation. In this paper we consider inferential validation relative to controllability; that is, if an inference procedure is applied to data generated from a gene regulatory network and an intervention procedure is designed on the inferred network, how well does it perform on the true network? The reasoning behind such a criterion is that, if our purpose is to use gene regulatory networks to design therapeutic intervention strategies, then we are not concerned with network fidelity, per se, but only with our ability to design effective interventions based on the inferred network. We will consider the problem from the perspectives of stationary control, which involves designing a control policy to be applied over time based on the current state of the network, with the decision procedure itself being time independent. The objective of a control policy is to optimally reduce the total steady-state probability mass of the undesirable states (phenotypes), which is equivalent to optimally increasing the total steady-state mass of the desirable states. Based on this criterion we compare several proposed network inference procedures. We will see that inference procedure ψ may perform poorer than inference procedure ξ relative to inferring the full network structure but perform better than ξ relative to controllability. Hence, when one is aiming at a specific application, it may be wise to use an objective-based measure of inference validity. No MeSH data available. Related in: MedlinePlus |
Related In:
Results -
Collection
License getmorefigures.php?uid=PMC3860259&req=5
Mentions: Finally, we evaluate different inference algorithms based on a metastatic melanoma network used in previous studies on network intervention (Qian and Dougherty, 2008; Qian et al., 2009; Yousefi and Dougherty, 2013). The network has 10 genes listed in the order from the most to the least significant bit: WNT5A, PIR, S100P, RET1, MMP3, PLCG1, MART1, HADHB, SNCA, and STC2. The order does not affect our analysis. We note here that this network was derived from gene expression data (Kim et al., 2002) collected in studies of metastatic melanoma (Bittner et al., 2000; Weeraratna et al., 2002). Table 2 and Figure 6 together illustrate the regulatory relationships among these selected 10 genes from 587 genes profiled in Bittner et al. (2000), Weeraratna et al. (2002), which were derived based on gene expression data rather than curated regulatory relationships among genes in literature. We believe that the model is appropriate for the purpose of illustrating the effectiveness of objective inferential validity on quantifying the performance of inference procedures in this work. Based on these information, we construct a BNp with the perturbation probability p = 0.01. As in the previous studies, the control objective is based on the fact that up-regulation of WNT5A is associated with increased metastasis. Thus, = {x/x1 = 1}. For this network, the undesirable steady-state mass is π = 0.2073 in the original network, which can be reduced as illustrated in Table 3 with different genes as potential targets using the MSSA algorithm on the original network. Based on this model, we simulate 20, 60, and 80 state transitions and infer the network based on these time series data using all five algorithms. As the primary objective here is to reduce the undesirable steady-state mass with WNT5A up-regulated, we focus on its shift derived by the MSSA algorithm based on the inferred networks using different inference algorithms. |
View Article: PubMed Central - PubMed
Affiliation: Department of Electrical and Computer Engineering, Texas A & M University TX, USA.
There are two distinct issues regarding network validation: (1) Does an inferred network provide good predictions relative to experimental data? (2) Does a network inference algorithm applied within a certain network model framework yield networks that are accurate relative to some criterion of goodness? The first issue concerns scientific validation and the second concerns algorithm validation. In this paper we consider inferential validation relative to controllability; that is, if an inference procedure is applied to data generated from a gene regulatory network and an intervention procedure is designed on the inferred network, how well does it perform on the true network? The reasoning behind such a criterion is that, if our purpose is to use gene regulatory networks to design therapeutic intervention strategies, then we are not concerned with network fidelity, per se, but only with our ability to design effective interventions based on the inferred network. We will consider the problem from the perspectives of stationary control, which involves designing a control policy to be applied over time based on the current state of the network, with the decision procedure itself being time independent. The objective of a control policy is to optimally reduce the total steady-state probability mass of the undesirable states (phenotypes), which is equivalent to optimally increasing the total steady-state mass of the desirable states. Based on this criterion we compare several proposed network inference procedures. We will see that inference procedure ψ may perform poorer than inference procedure ξ relative to inferring the full network structure but perform better than ξ relative to controllability. Hence, when one is aiming at a specific application, it may be wise to use an objective-based measure of inference validity.
No MeSH data available.