Limits...
A Sparse Reconstruction Approach for Identifying Gene Regulatory Networks Using Steady-State Experiment Data.

Zhang W, Zhou T - PLoS ONE (2015)

Bottom Line: Efficiency of this method is tested by an artificial linear network, a mitogen-activated protein kinase (MAPK) pathway network and the in silico networks of the DREAM challenges.The performance of the suggested approach is compared with two state-of-the-art algorithms, the widely adopted total least-squares (TLS) method and those available results on the DREAM project.Actual results show that, with a lower computational cost, the proposed method can significantly enhance estimation accuracy and greatly reduce false positive and negative errors.

View Article: PubMed Central - PubMed

Affiliation: School of Chemical Machinery, Qinghai University, Qinghai, China; Department of Automation, Tsinghua University, Beijing, China.

ABSTRACT

Motivation: Identifying gene regulatory networks (GRNs) which consist of a large number of interacting units has become a problem of paramount importance in systems biology. Situations exist extensively in which causal interacting relationships among these units are required to be reconstructed from measured expression data and other a priori information. Though numerous classical methods have been developed to unravel the interactions of GRNs, these methods either have higher computing complexities or have lower estimation accuracies. Note that great similarities exist between identification of genes that directly regulate a specific gene and a sparse vector reconstruction, which often relates to the determination of the number, location and magnitude of nonzero entries of an unknown vector by solving an underdetermined system of linear equations y = Φx. Based on these similarities, we propose a novel framework of sparse reconstruction to identify the structure of a GRN, so as to increase accuracy of causal regulation estimations, as well as to reduce their computational complexity.

Results: In this paper, a sparse reconstruction framework is proposed on basis of steady-state experiment data to identify GRN structure. Different from traditional methods, this approach is adopted which is well suitable for a large-scale underdetermined problem in inferring a sparse vector. We investigate how to combine the noisy steady-state experiment data and a sparse reconstruction algorithm to identify causal relationships. Efficiency of this method is tested by an artificial linear network, a mitogen-activated protein kinase (MAPK) pathway network and the in silico networks of the DREAM challenges. The performance of the suggested approach is compared with two state-of-the-art algorithms, the widely adopted total least-squares (TLS) method and those available results on the DREAM project. Actual results show that, with a lower computational cost, the proposed method can significantly enhance estimation accuracy and greatly reduce false positive and negative errors. Furthermore, numerical calculations demonstrate that the proposed algorithm may have faster convergence speed and smaller fluctuation than other methods when either estimate error or estimate bias is considered.

No MeSH data available.


Related in: MedlinePlus

Comparison of the ROC and PR curves in the DREAM4 identification using the SubLM1, SubLM2, TLS, SmOMP and StOMP algorithms.(a) ROC curves of Net2. (b) PR curves of Net2.
© Copyright Policy
Related In: Results  -  Collection

License
getmorefigures.php?uid=PMC4514654&req=5

pone.0130979.g005: Comparison of the ROC and PR curves in the DREAM4 identification using the SubLM1, SubLM2, TLS, SmOMP and StOMP algorithms.(a) ROC curves of Net2. (b) PR curves of Net2.

Mentions: We compare the SmOMP with the StOMP, SubLM1, SubLM2 and TLS algorithms for the DREAM3 and DREAM4 using only steady-state data. The corresponding ROC and PR curves of some typical estimations are respectively shown in Fig 4 for the Yeast2 in DREAM3, and Fig 5 for the Net2 in DREAM4. From these figures, it is obvious that the SmOMP algorithm is best among these five methods. Moreover, for every network of the DREAM3 and DREAM4 challenges, reconstruction results are respectively presented in Table 4. From these results and those available on the DREAM project website, we can conclude that the final score of proposed algorithm is much higher than Teams 296 which is top scorer among 22 participated teams in the DREAM3 challenge, and the estimation performances of the SmOMP algorithm significantly outperform Teams 236 which has been ranked the 14th place among 19 participated teams in the DREAM4 challenge. In addition, since our estimation procedures have significantly lower computational complexities, the SmOMP algorithm may be well appropriate and competent to identify large-scale GRNs. To be more specific, for the best of these challenges in DREAM3, it reported that 78h have been consumed to obtain an estimate a high-end cluster. However, utilizing a personal computer which is equipped with a 2.2 GHz CPU processor and a 2.0 GB RAM, SmOMP is required the averaged runtime 0.2730s, 0.5604s and 1.0538s for the 10-node, 50-node and 100-node network of the DREAM3 Ecoli1, respectively.


A Sparse Reconstruction Approach for Identifying Gene Regulatory Networks Using Steady-State Experiment Data.

Zhang W, Zhou T - PLoS ONE (2015)

Comparison of the ROC and PR curves in the DREAM4 identification using the SubLM1, SubLM2, TLS, SmOMP and StOMP algorithms.(a) ROC curves of Net2. (b) PR curves of Net2.
© Copyright Policy
Related In: Results  -  Collection

License
Show All Figures
getmorefigures.php?uid=PMC4514654&req=5

pone.0130979.g005: Comparison of the ROC and PR curves in the DREAM4 identification using the SubLM1, SubLM2, TLS, SmOMP and StOMP algorithms.(a) ROC curves of Net2. (b) PR curves of Net2.
Mentions: We compare the SmOMP with the StOMP, SubLM1, SubLM2 and TLS algorithms for the DREAM3 and DREAM4 using only steady-state data. The corresponding ROC and PR curves of some typical estimations are respectively shown in Fig 4 for the Yeast2 in DREAM3, and Fig 5 for the Net2 in DREAM4. From these figures, it is obvious that the SmOMP algorithm is best among these five methods. Moreover, for every network of the DREAM3 and DREAM4 challenges, reconstruction results are respectively presented in Table 4. From these results and those available on the DREAM project website, we can conclude that the final score of proposed algorithm is much higher than Teams 296 which is top scorer among 22 participated teams in the DREAM3 challenge, and the estimation performances of the SmOMP algorithm significantly outperform Teams 236 which has been ranked the 14th place among 19 participated teams in the DREAM4 challenge. In addition, since our estimation procedures have significantly lower computational complexities, the SmOMP algorithm may be well appropriate and competent to identify large-scale GRNs. To be more specific, for the best of these challenges in DREAM3, it reported that 78h have been consumed to obtain an estimate a high-end cluster. However, utilizing a personal computer which is equipped with a 2.2 GHz CPU processor and a 2.0 GB RAM, SmOMP is required the averaged runtime 0.2730s, 0.5604s and 1.0538s for the 10-node, 50-node and 100-node network of the DREAM3 Ecoli1, respectively.

Bottom Line: Efficiency of this method is tested by an artificial linear network, a mitogen-activated protein kinase (MAPK) pathway network and the in silico networks of the DREAM challenges.The performance of the suggested approach is compared with two state-of-the-art algorithms, the widely adopted total least-squares (TLS) method and those available results on the DREAM project.Actual results show that, with a lower computational cost, the proposed method can significantly enhance estimation accuracy and greatly reduce false positive and negative errors.

View Article: PubMed Central - PubMed

Affiliation: School of Chemical Machinery, Qinghai University, Qinghai, China; Department of Automation, Tsinghua University, Beijing, China.

ABSTRACT

Motivation: Identifying gene regulatory networks (GRNs) which consist of a large number of interacting units has become a problem of paramount importance in systems biology. Situations exist extensively in which causal interacting relationships among these units are required to be reconstructed from measured expression data and other a priori information. Though numerous classical methods have been developed to unravel the interactions of GRNs, these methods either have higher computing complexities or have lower estimation accuracies. Note that great similarities exist between identification of genes that directly regulate a specific gene and a sparse vector reconstruction, which often relates to the determination of the number, location and magnitude of nonzero entries of an unknown vector by solving an underdetermined system of linear equations y = Φx. Based on these similarities, we propose a novel framework of sparse reconstruction to identify the structure of a GRN, so as to increase accuracy of causal regulation estimations, as well as to reduce their computational complexity.

Results: In this paper, a sparse reconstruction framework is proposed on basis of steady-state experiment data to identify GRN structure. Different from traditional methods, this approach is adopted which is well suitable for a large-scale underdetermined problem in inferring a sparse vector. We investigate how to combine the noisy steady-state experiment data and a sparse reconstruction algorithm to identify causal relationships. Efficiency of this method is tested by an artificial linear network, a mitogen-activated protein kinase (MAPK) pathway network and the in silico networks of the DREAM challenges. The performance of the suggested approach is compared with two state-of-the-art algorithms, the widely adopted total least-squares (TLS) method and those available results on the DREAM project. Actual results show that, with a lower computational cost, the proposed method can significantly enhance estimation accuracy and greatly reduce false positive and negative errors. Furthermore, numerical calculations demonstrate that the proposed algorithm may have faster convergence speed and smaller fluctuation than other methods when either estimate error or estimate bias is considered.

No MeSH data available.


Related in: MedlinePlus