Genomics and medicine: an anticipation. From Boolean Mendelian genetics to multifactorial molecular medicine

Network-principled deep generative models for designing drug combinations as graph sets

Motivation

Combination therapy has shown to improve therapeutic efficacy while reducing side effects. Importantly, it has become an indispensable strategy to overcome resistance in antibiotics, antimicrobials and anticancer drugs. Facing enormous chemical space and unclear design principles for small-molecule combinations, computational drug-combination design has not seen generative models to meet its potential to accelerate resistance-overcoming drug combination discovery.

Results

We have developed the first deep generative model for drug combination design, by jointly embedding graph-structured domain knowledge and iteratively training a reinforcement learning-based chemical graph-set designer. First, we have developed hierarchical variational graph auto-encoders trained end-to-end to jointly embed gene–gene, gene–disease and disease–disease networks. Novel attentional pooling is introduced here for learning disease representations from associated genes’ representations. Second, targeting diseases in learned representations, we have recast the drug-combination design problem as graph-set generation and developed a deep learning-based model with novel rewards. Specifically, besides chemical validity rewards, we have introduced novel generative adversarial award, being generalized sliced Wasserstein, for chemically diverse molecules with distributions similar to known drugs. We have also designed a network principle-based reward for disease-specific drug combinations. Numerical results indicate that, compared to state-of-the-art graph embedding methods, hierarchical variational graph auto-encoder learns more informative and generalizable disease representations. Results also show that the deep generative models generate drug combinations following the principle across diseases. Case studies on four diseases show that network-principled drug combinations tend to have low toxicity. The generated drug combinations collectively cover the disease module similar to FDA-approved drug combinations and could potentially suggest novel systems pharmacology strategies. Our method allows for examining and following network-based principle or hypothesis to efficiently generate disease-specific drug combinations in a vast chemical combinatorial space.

Availability and implementation

https://github.com/Shen-Lab/Drug-Combo-Generator.

Supplementary information

Supplementary data are available at Bioinformatics online.

The pharmaceutical multi-activity of metallofullerenol invigorates cancer therapy

Currently, cancer continues to afflict humanity. The direct destruction and killing of tumor cells by surgery, radiation and chemotherapy gives rise to many side effects and compromised efficacy. Encouragingly, the rapid development of nanotechnology offers attractive opportunities to revolutionize the current situation of cancer therapy. Metallofullerenol Gd@C82(OH)22, in contrast to chemotherapeutics that directly kill tumor cells, demonstrates anti-tumor behavior with high efficiency and low toxicity by modulating the tumor microenvironment. Furthermore, Gd@C82(OH)22 has been recently reported to specifically target cancer stem cells. In this review, we give a concise introduction to the development of the fullerene family and then report the anti-tumor activity of Gd@C82(OH)22 based on its unique physicochemical characteristics, followed by a comprehensive summary of the anti-tumor biological mechanisms which target different components of the tumor microenvironment as well as the biodistribution and toxicity of Gd@C82(OH)22. Finally, we describe Gd@C82(OH)22 as a "particulate medicine" to highlight its distinctions from conventional "molecular medicine", with considerable emphasis on the advantages of nanomedicine. The in-depth investigation of Gd@C82(OH)22 undoubtedly provides a constructive reference for the development of other nanomedicines, especially in the fullerene family. The application of nanotechnology in the medical field definitely provides a promising and favorable future for improving the current status of cancer therapy.

Synthesis and Biological Activity of Isoflavone Derivatives from Chickpea as Potent Anti-Diabetic Agents

A set of novel isoflavone derivatives from chickpea were synthesized. The structures of derivatives were identified by proton nuclear magnetic resonance (¹H-NMR), carbon-13 (¹³C)-NMR and mass spectrometry (MS) spectral analyses. Their anti-diabetic activities were evaluated using an insulin-resistant (IR) HepG2 cell model. Additionally, the structure-activity relationships of these derivatives were briefly discussed. Compounds 1c, 2h, 3b, and 5 and genistein exhibited significant glucose consumption-enhancing effects in IR-HepG2 cells. In addition, the combinations of genistein, 2h, and 3b (combination 6) and of 3b, genistein, and 1c (combination 10) exhibited better anti-diabetic activity than the individual compounds. At the same dosage, there was no difference in effect between the combination 10 and the positive control (p > 0.05). Aditionally, we found the differences between the combination 10 and combination 6 for the protective effect of HUVEC (human umbilical vein endothelial cells) under high glucose concentration. The protective effects of combination 10 was stronger than combination 6, which suggested that combination 10 may have a better hypoglycemic activity in future studies. This study provides useful clues for the further design and discovery of anti-diabetic agents.

Multi-targeted therapy by curcumin: how spicy is it?

Although traditional medicines have been used for thousands of years, for most such medicines neither the active component nor their molecular targets have been very well identified. Curcumin, a yellow component of turmeric or curry powder, however, is an exception. Although inhibitors of cyclooxygenase-2, HER2, tumor necrosis factor, EGFR, Bcr-abl, proteosome, and vascular endothelial cell growth factor have been approved for human use by the United States Food and Drug Administration (FDA), curcumin as a single agent can down-regulate all these targets. Curcumin can also activate apoptosis, down-regulate cell survival gene products, and up-regulate p53, p21, and p27. Although curcumin is poorly absorbed after ingestion, multiple studies have suggested that even low levels of physiologically achievable concentrations of curcumin may be sufficient for its chemopreventive and chemotherapeutic activity. Thus, curcumin regulates multiple targets (multitargeted therapy), which is needed for treatment of most diseases, and it is inexpensive and has been found to be safe in human clinical trials. The present article reviews the key molecular mechanisms of curcumin action and compares this to some of the single-targeted therapies currently available for human cancer.

Salacia root, a unique Ayurvedic medicine, meets multiple targets in diabetes and obesity

In many traditional schools of medicine it is claimed that a balanced modulation of several targets can provide a superior therapeutic effect and decrease in side effect profile compared to a single action from a single selective ligand, especially in the treatment of certain chronic and complex diseases, such as diabetes and obesity. Diabetes and obesity have a multi-factorial basis involving both genetic and environmental risk factors. A wide array of medicinal plants and their active constituents play a role in the prevention and treatment of diabetes. Salacia roots have been used in Ayurvedic medicine for diabetes and obesity since antiquity, and have been extensively consumed in Japan, the United States and other countries as a food supplement for the prevention of obesity and diabetes. Recent pharmacological studies have demonstrated that Salacia roots modulate multiple targets: peroxisome proliferator-activated receptor-alpha-mediated lipogenic gene transcription, angiotensin II/angiotensin II type 1 receptor, alpha-glucosidase, aldose reductase and pancreatic lipase. These multi-target actions may mainly contribute to Salacia root-induced improvement of type 2 diabetes and obesity-associated hyperglycemia, dyslipidemia and related cardiovascular complications seen in humans and rodents. The results of bioassay-guided identification indicate that mangiferin, salacinol, kotalanol and kotalagenin 16-acetate are at least in part responsible for these multi-target regulatory activities of Salacia roots. The evidence suggests that this unique traditional medicine fulfills a multiple-target strategy in the prevention and treatment of diabetes and obesity. Although toxicological studies have suggested minimal adverse effects of the herbal medicine in rodents, a clinical trial is crucial to further confirm the safety of Salacia roots. In addition, further mechanistic studies are necessary in order to allow a better understanding of how use of Salacia root may interact with other therapeutic interventions.

Targeting cell signaling pathways for drug discovery: an old lock needs a new key

In this age of targeted therapy, the failure of most current drug-discovery efforts to yield safe, effective, and inexpensive drugs has generated widespread concern. Successful drug development has been stymied by a general focus on target selection rather than clinical safety and efficacy. The very process of validating the targets themselves is inefficient and in many cases leads to drugs having poor efficacy and undesirable side effects. Indeed, some rationally designed drugs (e.g., inhibitors of receptor tyrosine kinases, tumor necrosis factor (TNF), cyclooxygenase-2 (COX-2), vascular endothelial growth factor (VEGF), bcr-abl, and proteasomes) are ineffective against cancers and other inflammatory conditions and produce serious side effects. Since any given cancer carries mutations in an estimated 300 genes, this raises an important question about how effective these targeted therapies can ever be against cancer. Thus, it has become necessary to rethink drug development strategies. This review analyzes the shortcomings of rationally designed target-specific drugs against cancer cell signaling pathways and evaluates the available options for future drug development.

Multicomponent therapeutics for networked systems

Therapeutic regimens that comprise more than one active ingredient are commonly used in clinical medicine. Despite this, most drug discovery efforts search for drugs that are composed of a single chemical entity. A focus in the early drug discovery process on identifying and optimizing the activity of combinations of molecules can result in the identification of more effective drug regimens. A systems perspective facilitates an understanding of the mechanism of action of such drug combinations.

The drugs don't work: expectations and the shaping of pharmacogenetics

This article examines one particular set of technologies arising from developments in human genetics, those aimed at improving the targeting, design and use of conventional small molecule drugs-pharmacogenetics. Much of the debate about the applications and consequences of pharmacogenetics has been highly speculative, since little or no working technology is yet on the market. This article provides a novel analysis of the development of pharmacogenetics, and the social and ethical issues it raises, based on the sociology of technological expectations. In particular, it outlines how two alternative visions for the development of the technology are being articulated and embedded in a range of heterogeneous discourses, artefacts, actor strategies and practices, including: competing scientific research agendas, experimental technologies, emerging industrial structures and new ethical discourses. Expectations of how pharmacogenetics might emerge in each of these arenas are actively shaping the trajectory of this nascent technology and its potential socio-economic consequences.

Ethical reflections on pharmacogenetics and DNA banking in a cohort of HIV-infected patients

The aim of this study was to analyse ethical issues concerning the storage of human biological samples to be used in genetic analyses and pharmacogenetic research based on a French experience of DNA banking in a cohort of human immunodeficiency virus (HIV)-infected patients started on a protease inhibitor-containing treatment. We describe the ethical issues raised during the establishment of a DNA bank, including questions dealing with autonomy, benefit to the patient, information sharing and confidentiality as well as guarantees concerning the storage and use of DNA. The practical applications of themes illustrated theoretically in the literature are discussed. Most of the points raised are not specific to HIV, but some of them may be more accurate due to the characteristics of the HIV population, which is more involved in the social debate through the community life and the increased risk of stigmatization. Our results are summarized in the memorandum and consent form presented in the Appendices. One issue still open to discussion is the way the results of genetic data will be given to the patients. This work should allow other researchers and members of evaluation committees to enrich their considerations and should stimulate discussion on this topic.

Comparative genome analysis and pathway reconstruction

Pathway reconstruction builds on genome and biochemical data with the aim of reconstructing higher level interactions between identified enzymes in a specific genome, in particular the different enzyme pathways (species or individual/patient). Metabolite flow in a pathway is analyzed by different tools, such as elementary mode analysis. This reveals key enzymes and pharmacological targets in the enzyme network. An overview of bioinformatic tools and algorithms for these tasks, application examples and recent results from these techniques are presented. Target selection, drug development and optimization can all be sped up using these approaches.

Cambridge Healthtech Institute's Third Annual Conference on human genetic variation. 16-18 October 2000, Philadelphia, Pennsylvania, USA

A major goal of pharmacogenomics is to identify the human genetic variation that influences susceptibility to complex diseases. Recently, theoretical statistical analyses have suggested that genes for complex diseases may be found by linkage disequilibrium (i.e., association). Single nucleotide polymorphism (SNP) susceptibility alleles for common diseases can occur at high frequencies in various populations and, thus, have a major impact on morbidity and mortality. To be successful, SNP mapping studies require successful teamwork, integrating clinicians, epidemiologists, molecular genetics experts, laboratory automation engineers, bioinformatics and database experts. New statistical methods are also developing rapidly and promise to further increase the power of these studies. A recent conference on human genetic variation provided an opportunity for experts in all of these disciplines to exchange ideas. At present, great technological challenges need to be overcome in order to increase the throughput greatly while lowering cost and still maintaining high accuracy for SNP genotyping. Although this approach is relatively new (at least on the scale now being contemplated), the large payoffs anticipated to accrue from the successful mapping of SNPs in disease genes has led the area to be very strongly supported by both public and private funding sources. The potential payoff for improving disease diagnosis and therapeutic efficacy, with better avoidance of adverse events based on SNP associations, is providing a tremendous incentive to move this effort forward at an ever-accelerating pace.