Although lifestyle is an important driver for most – if not all – non-communicable diseases (NCDs),the underlying genetic component also contributes to disease etiology. Genetically predisposed individuals have an inherent risk that is independent of environmental factors. And unraveling the link between genetic variation and disease has become important, especially for developing new and more effective drugs against NCDs.

Early attempts to link genotype with phenotype was goaded by the expectation that even complex phenotypic traits would be governed by a handful of genetic loci. This initially spurred a large number of gene mapping studies, which was then followed by GWAS and, more recently, by exome sequencing. A surprising finding from these cumulative studies was the small effect sizes exerted by even the most important genetic loci, which could together explain only a small fraction of the predicted genetic variance.

This puzzle was subsequently rationalized by analyses revealing that the missing links in heritability could be accounted for by multiple single nucleotide polymorphisms (SNPs) with individual effect sizes that were below the threshold for statistical significance. The broad paradigm that emerged was that complex diseases are mediated through accumulation of weak effects on multiple genes, with consequent effects on regulatory pathways that drive disease risk.

The large number of causal genetic variants, each with tiny effect sizes on a phenotypic trait, has posed a serious challenge to understanding the relationship between genotype and phenotype. Indeed, recent results suggest that core genes contribute only a small part to total heritability whereas most other genes expressed in relevant cells likely also add through finite contributions.

A plausible explanation for this lies in the ‘small world’ architecture of cell regulatory networks, where any gene expressed in a disease-relevant tissue is connected through only just a few steps to the core genes. And since the total set of expressed genes will significantly outnumber the core genes, the sum of small effects across peripheral genes can far exceed contributions from genetic variations that directly affect the core genes themselves.

These observations were encapsulated by Pritchard and co-workers (1, 2) into the ‘omnigenic’ model of complex disease. This model proposes that essentially any gene with regulatory variants in at least one tissue that contributes to disease pathogenesis is likely to have non-trivial risk for that disease. Furthermore, since peripheral genes hugely outnumber core genes, a large fraction of the total genetic contribution to disease comes from peripheral genes that do not play direct roles in disease.

Implications for drug discovery.

The broad implications of the omnigenic model that disease-causing perturbations need not necessarily emanate from perturbations in onlythose genes (or proteins for that matter) that are present in the relevant biochemical pathways (core genes – or proteins), but can also derive from cumulative effects on peripheral genes or proteins, poses a serious challenge for the discovery of new drug targets. Traditionally, a good drug target is considered to be a protein that has a confirmed role in pathophysiology of the disease, and is disease modifying. Further, successful drug design also requires knowledge of the exact role the potential target plays in the disease.

An additional complication that derives from the networked architecture of cell regulatory mechanisms is that the ‘small world’ phenomenon confers functional redundancy to virtually all peripheral genes/proteins, and even many of the core genes/proteins. As a result, these networks exhibit nonlinear response behavior where neutralization of the disease-causing perturbation at a gene or protein (e.g. with a drug), need not necessarily reverse the disease phenotype.

This problem is especially relevant for cancers where a high degree of plasticity is associated with oncogenic pathway activation. Mutations in multiple signaling modules is often the norm in these cases, implying that the ideal targets for pathway inhibition need not necessarily coincide with those that are involved in its activation.

Plasticity and functional redundancy of cellular pathways, combined with non-linear response behavior of biological networks, pose the single most important hurdles to identification of new drug targets for NCDs today.This is particularly exemplified by genome-wide gene silencing screens, where the concordance of results from independent labs for the same target cell line is often very poor (<10%). Resolving this conundrum will require development of strategies that can distill through this complexity, and provide a coherent perspective on the core pathways involved.

From networks to core modules: localizing network vulnerability.

A potential approach to solving the problem of network complexity – and delineating candidate drug targets – was developed by one of our team members several years ago (3). This was during his studies on mitogenic signaling in cancer cells, where the goal was to identify vulnerable nodes in the mitogen-activated signaling network.

The approach was based on the recognition that the trade-off between sensitivity and robustness constrains the architecture of cell regulatory networks into a bow tie – or hourglass – structure. In the context of cell regulation, it was hypothesized that such a structure represents the convergence of diverse and redundant input processes onto a conserved core module of protein interactions.

Consistent with the tenets of information theory, it was postulated that these core elements function as key regulators of plasticity in the cellular response by calibrating the broad spectrum of inputs, into a defined set of output responses (3).

To test the hypothesis, an siRNA screen targeting the cellular signaling machinery was performed and the results obtained were subjected to graph theoretical analysis. An important feature of the latter analysis was the focus on reconstructing protein interaction networks through which effects of the identified signaling molecules – defined as input signals – were likely enforced on the output response of the cell cycle progression.

This approach eventually led to the identification of core modules that functioned as conserved regulatory elements that processed input signals, and translated them into the output response. Subsequent work revealed that these modules encapsulated mechanisms for coordinating seamless transition of cell through the individual cell cycle stages and, importantly, were functionally conserved across different cancer cell types.

Experimental validation of these findings could be achieved by pharmacological targeting of the least redundant nodes in these modules, which yielded disruption of the cell cycle in a tissue-type independent manner (3).

This study underscores that, although weak effects on multiple genes (or proteins) may collectively contribute to diseasetheir influences, nonetheless, likely converge onto a core module of protein-protein interactions. It is at such modules where their cumulative effects are integrated, and translated into the phenotypic output.

This discovery provides an opportunity for discovering new, and perhaps more effective, targets for drug development against NCDs. Especially significant in this context is the demonstration that delineation of convergent modules can help identify combinations of drug targets for achieving synergistic effects.


1. Boyle, E.A. et. al., Cell 169, 1177 – 1186, 2017

2. Liu, X. et. al., Cell 177, 1022 – 1034, 2019

3. Jailkhani, N. et. al., Genome Res. 21, 2067 – 2081, 2011

0 0 votes
Article Rating
Notify of
1 Comment
Newest Most Voted
Inline Feedbacks
View all comments
Rehana Shakeel
2 years ago

 Thank you for this terrific post with interesting links and for putting a spotlight on my Lead CRA blog.
Drug discovery