Limits...
How difficult is inference of mammalian causal gene regulatory networks?

Djordjevic D, Yang A, Zadoorian A, Rungrugeecharoen K, Ho JW - PLoS ONE (2014)

Bottom Line: Our data have thorough annotation of tissue types and embryonic stages, as well as the type of regulation (activation, inhibition and no effect), which uniquely allows us to estimate both sensitivity and specificity of the inference of tissue specific causal GRN edges.Using these unprecedented datasets, we found that gene co-expression does not reliably distinguish true positive from false positive interactions, making inference of GRN in mammalian development very difficult.Our result supports the importance of using perturbation experimental data in causal network reconstruction.

View Article: PubMed Central - PubMed

Affiliation: Victor Chang Cardiac Research Institute, Sydney, New South Wales, Australia; The University of New South Wales, Sydney, New South Wales, Australia.

ABSTRACT
Gene regulatory networks (GRNs) play a central role in systems biology, especially in the study of mammalian organ development. One key question remains largely unanswered: Is it possible to infer mammalian causal GRNs using observable gene co-expression patterns alone? We assembled two mouse GRN datasets (embryonic tooth and heart) and matching microarray gene expression profiles to systematically investigate the difficulties of mammalian causal GRN inference. The GRNs were assembled based on > 2,000 pieces of experimental genetic perturbation evidence from manually reading > 150 primary research articles. Each piece of perturbation evidence records the qualitative change of the expression of one gene following knock-down or over-expression of another gene. Our data have thorough annotation of tissue types and embryonic stages, as well as the type of regulation (activation, inhibition and no effect), which uniquely allows us to estimate both sensitivity and specificity of the inference of tissue specific causal GRN edges. Using these unprecedented datasets, we found that gene co-expression does not reliably distinguish true positive from false positive interactions, making inference of GRN in mammalian development very difficult. Nonetheless, if we have expression profiling data from genetic or molecular perturbation experiments, such as gene knock-out or signalling stimulation, it is possible to use the set of differentially expressed genes to recover causal regulatory relationships with good sensitivity and specificity. Our result supports the importance of using perturbation experimental data in causal network reconstruction. Furthermore, we showed that causal gene regulatory relationship can be highly cell type or developmental stage specific, suggesting the importance of employing expression profiles from homogeneous cell populations. This study provides essential datasets and empirical evidence to guide the development of new GRN inference methods for mammalian organ development.

Show MeSH

Related in: MedlinePlus

Fold changes (log) from tooth microarray perturbation experiments that matched the perturbation evidence in the literature show consistency with expected trends.RTPs that are inhibiting (A), have no effect (B), or are activating (C) trend to have negative, close to zero and positive fold changes respectively. (D) shows the consistency of the literature based RTP type (Lit.) and microarray data (M.A.) as fold change cut-off varies between 0 and 3 (both up- or down-regulation).
© Copyright Policy
Related In: Results  -  Collection

License
getmorefigures.php?uid=PMC4219746&req=5

pone-0111661-g004: Fold changes (log) from tooth microarray perturbation experiments that matched the perturbation evidence in the literature show consistency with expected trends.RTPs that are inhibiting (A), have no effect (B), or are activating (C) trend to have negative, close to zero and positive fold changes respectively. (D) shows the consistency of the literature based RTP type (Lit.) and microarray data (M.A.) as fold change cut-off varies between 0 and 3 (both up- or down-regulation).

Mentions: To examine whether the microarray data actually contain any information for identifying causal regulatory interactions, we investigated whether the set of differentially expressed genes from perturbation experiments can be used to infer causal regulatory relationships. The tooth microarray dataset contains 6 perturbation experiments, including transgenic knockdowns of Msx1 and Pax9, and exogenous stimulation of the BMP, Wnt, sonic hedgehog and FGF pathways. Based on the pathway information provided by [23], we identified 39 RTPs from the literature at stage E13 that corresponded to the microarray perturbation experiments. Encouragingly, the observed directions and fold changes of differential expression as determined by the microarray experiments were consistent with the regulatory relationships predicted by the literature (Figure 4). We found that using a fairly conservative absolute log fold change cut-off of 1 (i.e., 2-fold change) would result in a edge detection sensitivity (true positive rate) of 30%, increasing up to as the cut-off is relaxed. The false positive rate ( = 1-specificity) is consistently much lower than the true positive rate, suggesting that it is possible to distinguish causal gene regulation from non-regulatory ones with a reasonable sensitivity and specificity. We repeated the analysis considering all developmental stages, which increased the number of RTPs to 144. The trends are still visible although with increased noise (Figure S5).


How difficult is inference of mammalian causal gene regulatory networks?

Djordjevic D, Yang A, Zadoorian A, Rungrugeecharoen K, Ho JW - PLoS ONE (2014)

Fold changes (log) from tooth microarray perturbation experiments that matched the perturbation evidence in the literature show consistency with expected trends.RTPs that are inhibiting (A), have no effect (B), or are activating (C) trend to have negative, close to zero and positive fold changes respectively. (D) shows the consistency of the literature based RTP type (Lit.) and microarray data (M.A.) as fold change cut-off varies between 0 and 3 (both up- or down-regulation).
© Copyright Policy
Related In: Results  -  Collection

License
Show All Figures
getmorefigures.php?uid=PMC4219746&req=5

pone-0111661-g004: Fold changes (log) from tooth microarray perturbation experiments that matched the perturbation evidence in the literature show consistency with expected trends.RTPs that are inhibiting (A), have no effect (B), or are activating (C) trend to have negative, close to zero and positive fold changes respectively. (D) shows the consistency of the literature based RTP type (Lit.) and microarray data (M.A.) as fold change cut-off varies between 0 and 3 (both up- or down-regulation).
Mentions: To examine whether the microarray data actually contain any information for identifying causal regulatory interactions, we investigated whether the set of differentially expressed genes from perturbation experiments can be used to infer causal regulatory relationships. The tooth microarray dataset contains 6 perturbation experiments, including transgenic knockdowns of Msx1 and Pax9, and exogenous stimulation of the BMP, Wnt, sonic hedgehog and FGF pathways. Based on the pathway information provided by [23], we identified 39 RTPs from the literature at stage E13 that corresponded to the microarray perturbation experiments. Encouragingly, the observed directions and fold changes of differential expression as determined by the microarray experiments were consistent with the regulatory relationships predicted by the literature (Figure 4). We found that using a fairly conservative absolute log fold change cut-off of 1 (i.e., 2-fold change) would result in a edge detection sensitivity (true positive rate) of 30%, increasing up to as the cut-off is relaxed. The false positive rate ( = 1-specificity) is consistently much lower than the true positive rate, suggesting that it is possible to distinguish causal gene regulation from non-regulatory ones with a reasonable sensitivity and specificity. We repeated the analysis considering all developmental stages, which increased the number of RTPs to 144. The trends are still visible although with increased noise (Figure S5).

Bottom Line: Our data have thorough annotation of tissue types and embryonic stages, as well as the type of regulation (activation, inhibition and no effect), which uniquely allows us to estimate both sensitivity and specificity of the inference of tissue specific causal GRN edges.Using these unprecedented datasets, we found that gene co-expression does not reliably distinguish true positive from false positive interactions, making inference of GRN in mammalian development very difficult.Our result supports the importance of using perturbation experimental data in causal network reconstruction.

View Article: PubMed Central - PubMed

Affiliation: Victor Chang Cardiac Research Institute, Sydney, New South Wales, Australia; The University of New South Wales, Sydney, New South Wales, Australia.

ABSTRACT
Gene regulatory networks (GRNs) play a central role in systems biology, especially in the study of mammalian organ development. One key question remains largely unanswered: Is it possible to infer mammalian causal GRNs using observable gene co-expression patterns alone? We assembled two mouse GRN datasets (embryonic tooth and heart) and matching microarray gene expression profiles to systematically investigate the difficulties of mammalian causal GRN inference. The GRNs were assembled based on > 2,000 pieces of experimental genetic perturbation evidence from manually reading > 150 primary research articles. Each piece of perturbation evidence records the qualitative change of the expression of one gene following knock-down or over-expression of another gene. Our data have thorough annotation of tissue types and embryonic stages, as well as the type of regulation (activation, inhibition and no effect), which uniquely allows us to estimate both sensitivity and specificity of the inference of tissue specific causal GRN edges. Using these unprecedented datasets, we found that gene co-expression does not reliably distinguish true positive from false positive interactions, making inference of GRN in mammalian development very difficult. Nonetheless, if we have expression profiling data from genetic or molecular perturbation experiments, such as gene knock-out or signalling stimulation, it is possible to use the set of differentially expressed genes to recover causal regulatory relationships with good sensitivity and specificity. Our result supports the importance of using perturbation experimental data in causal network reconstruction. Furthermore, we showed that causal gene regulatory relationship can be highly cell type or developmental stage specific, suggesting the importance of employing expression profiles from homogeneous cell populations. This study provides essential datasets and empirical evidence to guide the development of new GRN inference methods for mammalian organ development.

Show MeSH
Related in: MedlinePlus