Diseases having multifactorial etiologies include cancer and cardiovascular disease, Alzheimer and Parkinson disease, diabetes and obesity, and depression and schizophrenia. These serious disorders are associated not only with multiple genes but also with multiple pathological mechanisms. Infectious diseases belong in this category because of the multicomponent character both of the immune response and of the metabolisms of bacterial pathogens. Nature herself provides compelling clues that multifactorial diseases require multifactorial treatments. For example, most tumorigenic viruses encode proteins that block multiple tumor-suppressor mechanisms in cells, and most hormones exert their physiological effects at many different sites simultaneously. Pharmacological clues include the well-known “wonder drug” aspirin, a single drug with multiple, mostly beneficial actions. Other pharmacological examples include the multi-drug treatments for diseases including AIDS, infection by drug-resistant bacteria, cancer, diabetes, and even some mood disorders that are by now standard. With a better understanding of multifactorial disease processes it should be possible to “repurpose” already approved drugs so that their combined “on-target” and “off-target” effects act synergistically to thwart specific disease processes. It is widely acknowledged that the main impediment to the design of multi-drug\multi-target treatments is the failure to understand the multifactorial processes themselves. New computational models are needed that can represent the interactions among the many factors involved, and new experimental methods are needed to evaluate the validity of the models. Fortunately, many of the required approaches are already being taken. Various bioinformatic and related “big-data” analyses are being used to predict the range of targets of approved drugs, and some of these predictions have been experimentally verified. Modeling modalities being brought to bear on the multifactoriality problem include static network analysis, kinetic models, flux balance analysis, systems of differential equations, Petri nets, Boolean and multivalued-logic networks, and rule-based models. Experimental approaches are based mainly on “high-throughput screens”, which can be used for automated assessment of many treatment combinations, each in multiple samples, in order to cover the range of possibilities and also ensure statistical power. A problem that plagues the development of pharmacological treatment strategies generally is the inability to reproduce relevant experimental results. This may be due mainly to the lack of standardization of preparations. In a perfect world, combinatorial experimental approaches would use standardized assays and would interact with computer modeling efforts such that model predictions are subjected to experimental verification, the results are used to confirm, correct, and extend the models, new predictions are generated and the cycle is repeated, yielding models of ever improving explanatory power. Such experimentally verified models could provide the understanding necessary to develop the multi-target/multi-drug treatments we need to combat the most serious diseases. The purpose of this Research Topic is to initiate a dialog between computational and experimental scientists in which problems are identified, solutions proposed, and movement toward a more perfect drug-development world is begun.
Diseases having multifactorial etiologies include cancer and cardiovascular disease, Alzheimer and Parkinson disease, diabetes and obesity, and depression and schizophrenia. These serious disorders are associated not only with multiple genes but also with multiple pathological mechanisms. Infectious diseases belong in this category because of the multicomponent character both of the immune response and of the metabolisms of bacterial pathogens. Nature herself provides compelling clues that multifactorial diseases require multifactorial treatments. For example, most tumorigenic viruses encode proteins that block multiple tumor-suppressor mechanisms in cells, and most hormones exert their physiological effects at many different sites simultaneously. Pharmacological clues include the well-known “wonder drug” aspirin, a single drug with multiple, mostly beneficial actions. Other pharmacological examples include the multi-drug treatments for diseases including AIDS, infection by drug-resistant bacteria, cancer, diabetes, and even some mood disorders that are by now standard. With a better understanding of multifactorial disease processes it should be possible to “repurpose” already approved drugs so that their combined “on-target” and “off-target” effects act synergistically to thwart specific disease processes. It is widely acknowledged that the main impediment to the design of multi-drug\multi-target treatments is the failure to understand the multifactorial processes themselves. New computational models are needed that can represent the interactions among the many factors involved, and new experimental methods are needed to evaluate the validity of the models. Fortunately, many of the required approaches are already being taken. Various bioinformatic and related “big-data” analyses are being used to predict the range of targets of approved drugs, and some of these predictions have been experimentally verified. Modeling modalities being brought to bear on the multifactoriality problem include static network analysis, kinetic models, flux balance analysis, systems of differential equations, Petri nets, Boolean and multivalued-logic networks, and rule-based models. Experimental approaches are based mainly on “high-throughput screens”, which can be used for automated assessment of many treatment combinations, each in multiple samples, in order to cover the range of possibilities and also ensure statistical power. A problem that plagues the development of pharmacological treatment strategies generally is the inability to reproduce relevant experimental results. This may be due mainly to the lack of standardization of preparations. In a perfect world, combinatorial experimental approaches would use standardized assays and would interact with computer modeling efforts such that model predictions are subjected to experimental verification, the results are used to confirm, correct, and extend the models, new predictions are generated and the cycle is repeated, yielding models of ever improving explanatory power. Such experimentally verified models could provide the understanding necessary to develop the multi-target/multi-drug treatments we need to combat the most serious diseases. The purpose of this Research Topic is to initiate a dialog between computational and experimental scientists in which problems are identified, solutions proposed, and movement toward a more perfect drug-development world is begun.