Network-based modular latent structure analysis
© Yu and Bai; licensee BioMed Central Ltd. 2014
Published: 13 November 2014
High-throughput expression data, such as gene expression and metabolomics data, exhibit modular structures. Groups of features in each module follow a latent factor model, while between modules, the latent factors are quasi-independent. Recovering the latent factors can shed light on the hidden regulation patterns of the expression. The difficulty in detecting such modules and recovering the latent factors lies in the high dimensionality of the data, and the lack of knowledge in module membership.
Here we describe a method based on community detection in the co-expression network. It consists of inference-based network construction, module detection, and interacting latent factor detection from modules.
In simulations, the method outperformed projection-based modular latent factor discovery when the input signals were not Gaussian. We also demonstrate the method's value in real data analysis.
The new method nMLSA (network-based modular latent structure analysis) is effective in detecting latent structures, and is easy to extend to non-linear cases. The method is available as R code at http://web1.sph.emory.edu/users/tyu8/nMLSA/.
Modularity is a common characteristic of high-throughput biological data . In a large system, the biological units, i.e. features (genes, proteins, or metabolites) are organized into quasi-autonomous modules. In expression data, each expression module can be modeled reasonably well using the latent factor approach [2, 3]. Given the involvement of thousands of features, an unknown number of modules, and unknown module membership of the features, it is difficult to faithfully detect the modules and recover the underlying latent factors controlling the modules.
Dimension reduction methods at the global level, such as Principal Component Analysis (PCA), Independent Component Analysis (ICA) , sparse PCA [5, 6], and Bayesian decomposition  are not effective in detecting localized signals. Clustering methods group co-expressed features together , which may help identify modules that are controlled by a single underlying signal [9, 10]. However in real data, the features involved in the same module may not co-express when more than one latent factors control the module. We previously proposed the projection-based Modular Latent Structure Analysis (MLSA) , which detects modules using iteratively re-weighted singular value decomposition (SVD). So far there are no other modular decomposition methods. In this study, we seek to improve the method using a totally different approach. Our goal is to develop a method that is more intuitive, flexible, and involves less ad hoc parameter choices.
Using networks constructed from expression data can provide a flexible framework for module detection [12–14]. Here we present a method to identify modules and the underlying latent signals in three steps: (1) constructing a co-expression network based on statistical inference and local false discovery rate (lfdr); (2) detecting communities in the network; and (3) recovering interacting latent factors from the modules.
The goal of the algorithm is to achieve modular matrix decomposition. We attempt to solve the problem by assembling tools from some well-established fields. The first is the reverse engineering of genome-scale networks. There are a number of methods available in this area, which were designed with different objectives, including Gaussian Graphical Models where the absence of an edge signifies conditional independence [15, 16], co-expression network where edges signify marginal dependence , information theory-based networks , and Bayesian networks . In this study, we designed our own method to estimate an inference-based co-expression network using the local false discovery rate (lfdr) concept [19–21]. The use of local fdr makes the procedure adaptive to shifts of baseline correlation levels and avoids constructing overly dense networks when there are pervasive low-level correlations between genes. Once the network is constructed, we borrow a method from the mature field of community detection in large networks [22–25]. This is followed by latent factor extraction and rotation using factor analysis methods . Added together, the assembled tools make a very good heuristic solution to the modular decomposition problem.
We demonstrate the superiority of the new method against existing modular and global decomposition methods using simulations, and apply the method to a real dataset to show it detects biologically meaningful modules that are controlled by multiple latent factors.
where q is the size of the module, r is the number of latent factors controlling the module, L is the regulation strength (loading) matrix, and E is the residual matrix. Our interest is estimating (1) the number of modules, (2) the module membership of the features, (3) the activities of the latent factors controlling each module (F matrix), and (4) the regulation strength of each factor on each feature (L matrix).
The estimation procedure
so that the distribution of the resulting statistic is close to normal under the null hypothesis that the pair of features are independent . Thirdly, we compute the local false discovery rate using Efron's procedure . The local fdr is a statistical statement of how likely two features are independent given we observe the statistics from all pairs of features. Fourth, if the local fdr value for a pair of features is smaller than a threshold, e.g. 0.2, an edge is established between the two features.
We then pool all the values computed from all pairs of communities and examine the distribution. Any outlier , defined by a value higher than the median plus four times the difference between the 75th percentile and the median, signifies a community pair that should be merged into a single community.
Step 3. Detecting latent factors from each module. For each module, we first conduct an eigenvalue decomposition of the covariance matrix, and select all eigenvectors that account for at least 5% of the data variance. We then find the projection length of each feature onto each eigenvector , where i denotes the feature and j denotes the eigenvector. The value is the number of features in the module, and is the number of eigenvectors under consideration.
Two eigenvectors are considered "interactive" if the correlation of the projection length of the features onto these two vectors is statistically significant. We initiate a selected vector set with only the first eigenvector. Then from the second eigenvector on, if the eigenvector is interactive with any vector in the selected set, it is added to the selected set. Otherwise we stop the iteration and return the selected vector set as the latent variables of the module. If more than one eigenvector is selected, we rotate them using oblique rotation .
The values in L can be filled in two ways. The first is by performing linear regression of each gene against only the factors of the modules the gene is assigned to. Alternatively, we can perform regularized regression of each gene against all the factors using lasso  with BIC (Bayesian information criterion) model selection.
The latent factor scores were either independent Gaussian, or randomly chosen from a mixture of four types: Gaussian, sine wave, square wave, and sawtooth wave (Additional file 1 Figure S1). The setting stayed the same for every module in each simulated dataset.
Different levels of within-module loading sparsity, i.e. proportion of zero loadings, were tested. The sparsity of the loading matrix was achieved by drawing samples from the binomial distribution. After the non-zero positions in the loading matrix was determined, for every simulated gene, if there were m controlling factors, we divided [0, 1]into m regions by drawing (m-1) samples from the uniform distribution between 0 and 1. We then used the sizes of the regions as the loadings for the gene. Half of the loadings were then multiplied by -1 to generate negative loadings. The sparsity levels tested were 0%, 30% and 60%. The setting stayed the same for every module in each simulated dataset.
After multiplying the loading matrix and the factor score matrix to generate the simulated expression matrix, Gaussian random noise was added to achieve different signal to noise ratios (values used: 1, 2). The setting stayed the same for every module in each simulated dataset.
The number of samples was set at 100. All possible combinations of the parameters were tested, each repeated 100 times.
To judge the performance of the methods, we used the information of the true hidden factors to group the identified factors. Let K be the combined hidden factor count from all modules in the simulated dataset. We first performed linear regression of every identified factor against each hidden factor group (those controlled the same module), and recorded the multiple R2. The identified factor was assigned to the hidden factor group with which it had the largest R2 value. The K identified factors with the largest R2 values were retained for the next step. Second, we performed linear regression of every true hidden factor against the identified factors assigned to its group, and recorded the multiple R2 as the level of recovery of the true hidden factor. The ideal method should yield multiple R2 values close to one. After repeating the simulation from every parameter setting 100 times, we compared the methods by the distribution of the multiple R2 values.
The simulation results are summarized in Figure 2. Each sub-plot represents a parameter setting. The relative frequencies (10 equal-sized bins between 0 and 1, equivalent to the histogram) of the R2 values are plotted in Figure 1. Different colors represent different methods. The curves are effectively histograms of the multiple R2 values. The curve of a better method should show higher frequency in larger R2 values. In all the scenarios, clearly nMLSA (red) and MLSA (blue) outperformed the other methods.
Real data analysis
Top 25 GO terms overrepresented by genes associated with factor 1.
ribonucleoprotein complex biogenesis
ncRNA metabolic process
rRNA metabolic process
cellular component biogenesis at cellular level
ribosomal large subunit biogenesis
maturation of 5.8S rRNA
maturation of 5.8S rRNA from tricistronic rRNA transcript (SSU-rRNA, 5.8S rRNA, LSU-rRNA)
RNA metabolic process
ribosomal small subunit biogenesis
cellular component biogenesis
maturation of SSU-rRNA
maturation of SSU-rRNA from tricistronic rRNA transcript (SSU-rRNA, 5.8S rRNA, LSU-rRNA)
cleavage involved in rRNA processing
nucleic acid metabolic process
endonucleolytic cleavage involved in rRNA processing
endonucleolytic cleavage of tricistronic rRNA transcript (SSU-rRNA, 5.8S rRNA, LSU-rRNA)
nucleobase-containing compound metabolic process
endonucleolytic cleavage in ITS1 to separate SSU-rRNA from 5.8S rRNA and LSU-rRNA from tricistronic rRNA transcript (SSU-rRNA, 5.8S rRNA, LSU-rRNA)
endonucleolytic cleavage to generate mature 5'-end of SSU-rRNA from (SSU-rRNA, 5.8S rRNA, LSU-rRNA)
Top 25 GO terms overrepresented by genes associated with factor 2.
ubiquitin-dependent protein catabolic process
modification-dependent protein catabolic process
cellular protein catabolic process
proteolysis involved in cellular protein catabolic process
modification-dependent macromolecule catabolic process
protein catabolic process
proteasomal ubiquitin-independent protein catabolic process
cellular macromolecule catabolic process
cellular protein complex assembly
macromolecule catabolic process
cellular component organization at cellular level
proteasomal ubiquitin-dependent protein catabolic process
establishment of protein localization
proteasomal protein catabolic process
protein complex assembly
cellular protein localization
protein complex biogenesis
cellular macromolecule localization
cellular macromolecular complex subunit organization
Top 25 GO terms overrepresented by genes associated with factor 3.
mitotic cell cycle
cell cycle process
cell cycle phase
microtubule cytoskeleton organization
M phase of mitotic cell cycle
mitotic spindle organization in nucleus
regulation of microtubule-based process
regulation of microtubule cytoskeleton organization
cellular component organization
interphase of mitotic cell cycle
mitotic spindle organization
cellular component movement
regulation of cell cycle process
Top 25 GO terms overrepresented by genes associated with factor 4.
mitotic cell cycle
M phase of mitotic cell cycle
cell cycle process
cell cycle phase
exit from mitosis
cell cycle cytokinesis
nuclear migration along microtubule
cytoskeleton-dependent intracellular transport
organelle transport along microtubule
regulation of transcription involved in G1 phase of mitotic cell cycle
short-chain fatty acid metabolic process
DNA geometric change
positive regulation of spindle pole body separation
In this study, we developed the network-based modular latent structure analysis (nMLSA). It is aimed at detecting expression modules and latent factors controlling the modules, the same goal as the original MLSA . Compared to MLSA, the new method is based on a totally different setup, and is substantially advantageous. Firstly, the number of tuning parameters and heuristic choices is substantially less compared to MLSA. Secondly, the method is much more intuitive to understand. Thirdly, it is more flexible. As an example, one can easily limit the gene relations to positive correlations and ignore negative correlations using nMLSA, while MLSA has to take both positive and negative correlations. Fourth, nMLSA can be adapted for nonlinearly associated modules if a nonlinear association measure is used in the co-expression network building, while MLSA is limited to linear relations. In the nonlinear case, it is difficult to define latent factors. The challenge is subject to our future studies.
Instead of using hard cutoffs, nMLSA utilizes the concept of local false discovery rate (lfdr). As different datasets exhibit different levels of baseline correlation , using hard cutoffs on correlations may result in unsatisfactory results. Using local false discovery rate procedures that are flexible in the null distribution estimation, nMLSA is naturally adaptive to the characteristics of the data. Given the nMLSA procedure relies on existing network community detection algorithms, it is admitted that the performance of the method relies on the choice of the community detection algorithm. The research field of community detection is mature and a number of good methods are available. Thus it is not difficult to tune the method to achieve good performance.
In summary, the new network-based method nMLSA is more effective than existing methods in recovering biologically meaningful latent variables and latent variable groups. The method can potentially be extended to detect nonlinearly associated modules if a nonlinear association measure is used to build the network.
This work was partially supported by NIH grants P20HL113451, P30AI50409 and U19AI090023. The funding source to publish the publication cost is NIH grant U19AI090023.
This article has been published as part of BMC Bioinformatics Volume 15 Supplement 13, 2014: Selected articles from the 9th International Symposium on Bioinformatics Research and Applications (ISBRA'13): Bioinformatics. The full contents of the supplement are available online at http://www.biomedcentral.com/bmcbioinformatics/supplements/15/S13.
- Wagner GP, Pavlicev M, Cheverud JM: The road to modularity. Nat Rev Genet. 2007, 8: (12):921-931.PubMedGoogle Scholar
- Yu T, Li KC: Inference of transcriptional regulatory network by two-stage constrained space factor analysis. Bioinformatics. 2005, 21 (21): 4033-4038. 10.1093/bioinformatics/bti656.View ArticlePubMedGoogle Scholar
- Liao JC, Boscolo R, Yang YL, Tran LM, Sabatti C, Roychowdhury VP: Network component analysis: reconstruction of regulatory signals in biological systems. Proc Natl Acad Sci USA. 2003, 100 (26): 15522-15527. 10.1073/pnas.2136632100.PubMed CentralView ArticlePubMedGoogle Scholar
- Lee TW: Independent component analysis : theory and applications. 1998, Boston: Kluwer Academic PublishersGoogle Scholar
- Zou H, Hastie T, Tibshirani R: Sparse principal component analysis. Journal of Computational and Graphical Statistics. 2006, 15 (2): 265-286. 10.1198/106186006X113430.View ArticleGoogle Scholar
- Carvalho CM, Chang J, Lucas JE, Nevins JR, Wang Q, West M: High-Dimensional Sparse Factor Modeling: Applications in Gene Expression Genomics. Journal of the American Statistical Association. 2008, 103 (484): 1438-1456. 10.1198/016214508000000869.PubMed CentralView ArticlePubMedGoogle Scholar
- Moloshok TD, Klevecz RR, Grant JD, Manion FJ, Speier WFt, Ochs MF: Application of Bayesian decomposition for analysing microarray data. Bioinformatics. 2002, 18 (4): 566-575. 10.1093/bioinformatics/18.4.566.View ArticlePubMedGoogle Scholar
- Gan G, Ma C, Wu J: Data clustering : theory, algorithms, and applications. 2007, Philadelphia, Pa. Alexandria, Va.: SIAM. American Statistical AssociationView ArticleGoogle Scholar
- Hastie T, Tibshirani R, Eisen MB, Alizadeh A, Levy R, Staudt L, Chan WC, Botstein D, Brown P: 'Gene shaving' as a method for identifying distinct sets of genes with similar expression patterns. Genome Biol. 2000, 1 (2): RESEARCH0003-PubMed CentralView ArticlePubMedGoogle Scholar
- Yuan S, Li KC: Context-dependent clustering for dynamic cellular state modeling of microarray gene expression. Bioinformatics. 2007, 23 (22): 3039-3047. 10.1093/bioinformatics/btm457.View ArticlePubMedGoogle Scholar
- Yu T: An exploratory data analysis method to reveal modular latent structures in high-throughput data. BMC Bioinformatics. 2010, 11: 440-10.1186/1471-2105-11-440.PubMed CentralView ArticlePubMedGoogle Scholar
- Fu Q, Lemmens K, Sanchez-Rodreiquez A, Thijs IM, Meysman P, Sun H, Fierro AC, Engelen K, Marchal K: Directed module detection in a large-scale expression compendium. Methods Mol Biol. 2012, 804: 131-165. 10.1007/978-1-61779-361-5_8.View ArticlePubMedGoogle Scholar
- Langfelder P, Horvath S: WGCNA: an R package for weighted correlation network analysis. BMC Bioinformatics. 2008, 9: 559-10.1186/1471-2105-9-559.PubMed CentralView ArticlePubMedGoogle Scholar
- Li A, Horvath S: Network module detection: Affinity search technique with the multi-node topological overlap measure. BMC Res Notes. 2009, 2: 142-10.1186/1756-0500-2-142.PubMed CentralView ArticlePubMedGoogle Scholar
- Peng J, Wang P, Zhou N, Zhu J: Partial Correlation Estimation by Joint Sparse Regression Models. J Am Stat Assoc. 2009, 104 (486): 735-746. 10.1198/jasa.2009.0126.PubMed CentralView ArticlePubMedGoogle Scholar
- Schafer J, Strimmer K: An empirical Bayes approach to inferring large-scale gene association networks. Bioinformatics. 2005, 21 (6): 754-764. 10.1093/bioinformatics/bti062.View ArticlePubMedGoogle Scholar
- Margolin AA, Nemenman I, Basso K, Wiggins C, Stolovitzky G, Dalla Favera R, Califano A: ARACNE: an algorithm for the reconstruction of gene regulatory networks in a mammalian cellular context. BMC Bioinformatics. 2006, 7 (Suppl 1): S7-10.1186/1471-2105-7-S1-S7.PubMed CentralView ArticlePubMedGoogle Scholar
- Ellis B, Wong W: Learning causal bayesian network structures from experimental data. Journal of the American Statistical Association. 2008, 103: 778-789. 10.1198/016214508000000193.View ArticleGoogle Scholar
- Efron B, Tibshirani R: Empirical Bayes methods and false discovery rates for microarrays. Genet Epidemiol. 2002, 23 (1): 70-86. 10.1002/gepi.1124.View ArticlePubMedGoogle Scholar
- Strimmer K: A unified approach to false discovery rate estimation. BMC Bioinformatics. 2008, 9: 303-10.1186/1471-2105-9-303.PubMed CentralView ArticlePubMedGoogle Scholar
- Guedj M, Robin S, Celisse A, Nuel G: Kerfdr: a semi-parametric kernel-based approach to local false discovery rate estimation. BMC Bioinformatics. 2009, 10: 84-10.1186/1471-2105-10-84.PubMed CentralView ArticlePubMedGoogle Scholar
- Ball B, Karrer B, Newman ME: Efficient and principled method for detecting communities in networks. Physical review E, Statistical, nonlinear, and soft matter physics. 2011, 84 (3 Pt 2): 036103-View ArticlePubMedGoogle Scholar
- Hofman JM, Wiggins CH: Bayesian approach to network modularity. Physical review letters. 2008, 100 (25): 258701-PubMed CentralView ArticlePubMedGoogle Scholar
- Newman ME: Spectral methods for community detection and graph partitioning. Physical review E, Statistical, nonlinear, and soft matter physics. 2013, 88 (4): 042822-View ArticlePubMedGoogle Scholar
- Pons P, Latapy M: Computing Communities in Large Networks Using Random Walks. Lecture Notes in Computer Science. 2005, 3733: 284-293. 10.1007/11569596_31.View ArticleGoogle Scholar
- Bernaards CA, Jennrich RI: Gradient Projection Algorithms and Software for Arbitrary Rotation Criteria in Factor Analysis. Educational and Psychological Measurement. 2005, 65: 676-696. 10.1177/0013164404272507.View ArticleGoogle Scholar
- Rahman NA: A course in theoretical statistics for sixth forms, technical colleges, colleges of education, universities. 1968, London,: GriffinGoogle Scholar
- Efron B, Hastie T, Johnstone I, Tibshirani R: Least Angle Regression. Annals of Statistics. 2003, 32 (2): 407-499.Google Scholar
- Hastie T, Tibshirani R, Friedman JH: The elements of statistical learning : data mining, inference, and prediction. 2009, New York, NY: Springer, 2View ArticleGoogle Scholar
- Spellman PT, Sherlock G, Zhang MQ, Iyer VR, Anders K, Eisen MB, Brown PO, Botstein D, Futcher B: Comprehensive identification of cell cycle-regulated genes of the yeast Saccharomyces cerevisiae by microarray hybridization. Mol Biol Cell. 1998, 9 (12): 3273-3297. 10.1091/mbc.9.12.3273.PubMed CentralView ArticlePubMedGoogle Scholar
- Li KC, Yan M, Yuan SS: A simple statistical model for depicting the cdc15-synchronized yeast cell-cycle regulated gene expression data. Stat Sinica. 2002, 12 (1): 141-158.Google Scholar
- Ashburner M, Ball CA, Blake JA, Botstein D, Butler H, Cherry JM, Davis AP, Dolinski K, Dwight SS, Eppig JT: Gene ontology: tool for the unification of biology. The Gene Ontology Consortium. Nat Genet. 2000, 25 (1): 25-29. 10.1038/75556.PubMed CentralView ArticlePubMedGoogle Scholar
- Falcon S, Gentleman R: Using GOstats to test gene lists for GO term association. Bioinformatics. 2007, 23 (2): 257-258. 10.1093/bioinformatics/btl567.View ArticlePubMedGoogle Scholar
- Dai CL, Shi J, Chen Y, Iqbal K, Liu F, Gong CX: Inhibition of protein synthesis alters protein degradation through activation of protein kinase B (AKT). J Biol Chem. 2013Google Scholar
- Hemmerlin A: Post-translational events and modifications regulating plant enzymes involved in isoprenoid precursor biosynthesis. Plant Sci. 2013, 203-204: 41-54.View ArticlePubMedGoogle Scholar
- Stein H, Honig A, Miller G, Erster O, Eilenberg H, Csonka LN, Szabados L, Koncz C, Zilberstein A: Elevation of free proline and proline-rich protein levels by simultaneous manipulations of proline biosynthesis and degradation in plants. Plant Sci. 2011, 181 (2): 140-150. 10.1016/j.plantsci.2011.04.013.View ArticlePubMedGoogle Scholar
- Yu T, Zhao Y, Shen S: AAPL: Assessing Association between P-value Lists. Statistical analysis and data mining. 2013, 6 (2): 144-155. 10.1002/sam.11180.PubMed CentralView ArticlePubMedGoogle Scholar
This article is published under license to BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/4.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited. The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated.