Limits...
A structure-based approach for mapping adverse drug reactions to the perturbation of underlying biological pathways.

Wallach I, Jaitly N, Lilien R - PLoS ONE (2010)

Bottom Line: Understanding the biological processes behind the occurrence of ADRs would lead to the development of safer and more effective drugs.As proof-of-concept, we examined a dataset of 506 ADRs, 730 drugs, and 830 human protein targets.We found 32 associations to be supported by the scientific literature.

View Article: PubMed Central - PubMed

Affiliation: Department of Computer Science, University of Toronto, Toronto, Ontario, Canada. izharw@cs.toronto.edu

ABSTRACT
Adverse drug reactions (ADR), also known as side-effects, are complex undesired physiologic phenomena observed secondary to the administration of pharmaceuticals. Several phenomena underlie the emergence of each ADR; however, a dominant factor is the drug's ability to modulate one or more biological pathways. Understanding the biological processes behind the occurrence of ADRs would lead to the development of safer and more effective drugs. At present, no method exists to discover these ADR-pathway associations. In this paper we introduce a computational framework for identifying a subset of these associations based on the assumption that drugs capable of modulating the same pathway may induce similar ADRs. Our model exploits multiple information resources. First, we utilize a publicly available dataset pairing drugs with their observed ADRs. Second, we identify putative protein targets for each drug using the protein structure database and in-silico virtual docking. Third, we label each protein target with its known involvement in one or more biological pathways. Finally, the relationships among these information sources are mined using multiple stages of logistic-regression while controlling for over-fitting and multiple-hypothesis testing. As proof-of-concept, we examined a dataset of 506 ADRs, 730 drugs, and 830 human protein targets. Our method yielded 185 ADR-pathway associations of which 45 were selected to undergo a manual literature review. We found 32 associations to be supported by the scientific literature.

Show MeSH

Related in: MedlinePlus

An illustration of the network of pathway-ADR associations inferred by our model.Side-effects are represented as red squares and pathways as blue circles. The full list of 185 associations is available at Table S3. The 22 associations most strongly supported by the literature are circled. Pathways are colored by their KEGG categories.
© Copyright Policy
Related In: Results  -  Collection


getmorefigures.php?uid=PMC2925884&req=5

pone-0012063-g002: An illustration of the network of pathway-ADR associations inferred by our model.Side-effects are represented as red squares and pathways as blue circles. The full list of 185 associations is available at Table S3. The 22 associations most strongly supported by the literature are circled. Pathways are colored by their KEGG categories.

Mentions: We developed a computational model that associates ADRs with biological pathways. The model is summarized in this paragraph (full details appear in the Methods section). We used the empirically observed relations between drugs and ADRs, the mapping of target proteins into pathways, and the virtual normalized docking results of drugs onto experimentally determined human protein structures to construct a graphical representation of these relations (Figure 1). The current study utilized 730 small-molecule clinically-approved drugs [15] and 830 target proteins with solved structures [17]. The protein targets were associated with 176 human related pathways extracted from the KEGG database [16]. The drugs were associated with 506 ADRs extracted from the SIDER database [15]. Our computational model learns a set of connections from ADRs to pathways consistent with the observed drug-ADR relations. We note that for some drugs, the ADRs listed in the SIDER database are not traditional ADRs. For example, viral and bacterial infections, such as herpes and tuberculosis, are listed as ADRs but clearly, do not result from direct perturbations of a biological pathway. These ADRs reflect the observations of viral/bacterial infections while patients were under the administration of corresponding drugs. Viral infections may be listed as ADR if a drug perturbs some biological mechanisms and, as a result, increases vulnerability to viral infections. Alternatively, an ADR may not reflect a causative relation. For example, cancer or HIV patients having a stressed immune system, may suffer from increased vulnerability to infections. Consequently, cancer or HIV related drugs may be associated with infectious related ADRs without any causative underlying mechanism. In the results below we demonstrate several such examples. For clarity, we will refer to these biological pathway-ADR associations as simply associations. The learning of these associations employed a 2-phase logistic-regression model with L1-regularization and feature selection [18], [19]. This method was designed to avoid overfitting the small number of input samples by using techniques that heavily penalize complex models. As a result, the model identifies a small informative set of associations for which there is the most evidence. Utilizing a multiple testing correction with a false-discovery rate of 2% our model identified a set of 185 associations worthy of further investigation (Figure 2 and Table S3). These 185 associations involve 90 pathways and 121 ADRs and represent an extremely small fraction of the 89,056 possible pathway-ADR associations (176 pathways and 506 ADRs).


A structure-based approach for mapping adverse drug reactions to the perturbation of underlying biological pathways.

Wallach I, Jaitly N, Lilien R - PLoS ONE (2010)

An illustration of the network of pathway-ADR associations inferred by our model.Side-effects are represented as red squares and pathways as blue circles. The full list of 185 associations is available at Table S3. The 22 associations most strongly supported by the literature are circled. Pathways are colored by their KEGG categories.
© Copyright Policy
Related In: Results  -  Collection

Show All Figures
getmorefigures.php?uid=PMC2925884&req=5

pone-0012063-g002: An illustration of the network of pathway-ADR associations inferred by our model.Side-effects are represented as red squares and pathways as blue circles. The full list of 185 associations is available at Table S3. The 22 associations most strongly supported by the literature are circled. Pathways are colored by their KEGG categories.
Mentions: We developed a computational model that associates ADRs with biological pathways. The model is summarized in this paragraph (full details appear in the Methods section). We used the empirically observed relations between drugs and ADRs, the mapping of target proteins into pathways, and the virtual normalized docking results of drugs onto experimentally determined human protein structures to construct a graphical representation of these relations (Figure 1). The current study utilized 730 small-molecule clinically-approved drugs [15] and 830 target proteins with solved structures [17]. The protein targets were associated with 176 human related pathways extracted from the KEGG database [16]. The drugs were associated with 506 ADRs extracted from the SIDER database [15]. Our computational model learns a set of connections from ADRs to pathways consistent with the observed drug-ADR relations. We note that for some drugs, the ADRs listed in the SIDER database are not traditional ADRs. For example, viral and bacterial infections, such as herpes and tuberculosis, are listed as ADRs but clearly, do not result from direct perturbations of a biological pathway. These ADRs reflect the observations of viral/bacterial infections while patients were under the administration of corresponding drugs. Viral infections may be listed as ADR if a drug perturbs some biological mechanisms and, as a result, increases vulnerability to viral infections. Alternatively, an ADR may not reflect a causative relation. For example, cancer or HIV patients having a stressed immune system, may suffer from increased vulnerability to infections. Consequently, cancer or HIV related drugs may be associated with infectious related ADRs without any causative underlying mechanism. In the results below we demonstrate several such examples. For clarity, we will refer to these biological pathway-ADR associations as simply associations. The learning of these associations employed a 2-phase logistic-regression model with L1-regularization and feature selection [18], [19]. This method was designed to avoid overfitting the small number of input samples by using techniques that heavily penalize complex models. As a result, the model identifies a small informative set of associations for which there is the most evidence. Utilizing a multiple testing correction with a false-discovery rate of 2% our model identified a set of 185 associations worthy of further investigation (Figure 2 and Table S3). These 185 associations involve 90 pathways and 121 ADRs and represent an extremely small fraction of the 89,056 possible pathway-ADR associations (176 pathways and 506 ADRs).

Bottom Line: Understanding the biological processes behind the occurrence of ADRs would lead to the development of safer and more effective drugs.As proof-of-concept, we examined a dataset of 506 ADRs, 730 drugs, and 830 human protein targets.We found 32 associations to be supported by the scientific literature.

View Article: PubMed Central - PubMed

Affiliation: Department of Computer Science, University of Toronto, Toronto, Ontario, Canada. izharw@cs.toronto.edu

ABSTRACT
Adverse drug reactions (ADR), also known as side-effects, are complex undesired physiologic phenomena observed secondary to the administration of pharmaceuticals. Several phenomena underlie the emergence of each ADR; however, a dominant factor is the drug's ability to modulate one or more biological pathways. Understanding the biological processes behind the occurrence of ADRs would lead to the development of safer and more effective drugs. At present, no method exists to discover these ADR-pathway associations. In this paper we introduce a computational framework for identifying a subset of these associations based on the assumption that drugs capable of modulating the same pathway may induce similar ADRs. Our model exploits multiple information resources. First, we utilize a publicly available dataset pairing drugs with their observed ADRs. Second, we identify putative protein targets for each drug using the protein structure database and in-silico virtual docking. Third, we label each protein target with its known involvement in one or more biological pathways. Finally, the relationships among these information sources are mined using multiple stages of logistic-regression while controlling for over-fitting and multiple-hypothesis testing. As proof-of-concept, we examined a dataset of 506 ADRs, 730 drugs, and 830 human protein targets. Our method yielded 185 ADR-pathway associations of which 45 were selected to undergo a manual literature review. We found 32 associations to be supported by the scientific literature.

Show MeSH
Related in: MedlinePlus