Benefiting from big data in natural products: importance of preserving foundational skills and prioritizing data quality

Artificial intelligence for natural product drug discovery

Developments in computational omics technologies have provided new means to access the hidden diversity of natural products, unearthing new potential for drug discovery. In parallel, artificial intelligence approaches such as machine learning have led to exciting developments in the computational drug design field, facilitating biological activity prediction and de novo drug design for molecular targets of interest. Here, we describe current and future synergies between these developments to effectively identify drug candidates from the plethora of molecules produced by nature. We also discuss how to address key challenges in realizing the potential of these synergies, such as the need for high-quality datasets to train deep learning algorithms and appropriate strategies for algorithm validation.

Marine indole alkaloid diversity and bioactivity. What do we know and what are we missing?

Covering: marine indole alkaloids (n = 2048) and their reported bioactivities up to the end of 2021Despite increasing numbers of marine natural products (MNPs) reported each year, most have only been examined for cytotoxic, antibacterial, and/or antifungal biological activities with the majority found to be inactive in these assays. In this context, why are natural products continuing to be examined in assays they are unlikely to show significant activity in, and what targets might be more useful for expanding knowledge of their biologically relevant chemical space? We have undertaken a meta-analysis of the biological activities for 2048 marine indole alkaloids (MIAs), a diverse sub-class of MNPs reported up to the end of 2021, and this has highlighted that the bioactivity potentials for up to 86% of published MIAs remains underexplored and/or undefined. Although most published MIAs are not cytotoxic or antimicrobial, there is a continued focus on using these assays to evaluate new structurally related analogues. Using cheminformatics analyses, the chemical diversity of the 2048 MIAs were clustered using fragment based fingerprints and their reported bioactivity potency towards specific disease targets was assessed for structure activity trends. These analyses showed that there are groups of MIAs that possess potent and diverse activities and that many analogues, previously tested only in cellular toxicity assays, could be better exploited to generate structure activity relationships associated with leads to treat emerging diseases. A collection of indole drug and drug-lead structures from non-natural sources were also incorporated into the dataset providing complementary bioactivity profiles that were further used to predict underexplored areas of potential new activity and to better direct future testing of MIAs. Our findings clearly suggest the biological evaluation of MIAs continues to be conducted on a narrow range of bioassays and disease targets, and that shifting the focus to non-toxic disease targets should provide expanded knowledge of biologically relevant chemical space aimed at maximising the potential of MIAs for drug discovery.

Open data and algorithms for open science in AI-driven molecular informatics

Recent years have seen a sharp increase in the development of deep learning and artificial intelligence-based molecular informatics. There has been a growing interest in applying deep learning to several subfields, including the digital transformation of synthetic chemistry, extraction of chemical information from the scientific literature, and AI in natural product-based drug discovery. The application of AI to molecular informatics is still constrained by the fact that most of the data used for training and testing deep learning models are not available as FAIR and open data. As open science practices continue to grow in popularity, initiatives which support FAIR and open data as well as open-source software have emerged. It is becoming increasingly important for researchers in the field of molecular informatics to embrace open science and to submit data and software in open repositories. With the advent of open-source deep learning frameworks and cloud computing platforms, academic researchers are now able to deploy and test their own deep learning models with ease. With the development of new and faster hardware for deep learning and the increasing number of initiatives towards digital research data management infrastructures, as well as a culture promoting open data, open source, and open science, AI-driven molecular informatics will continue to grow. This review examines the current state of open data and open algorithms in molecular informatics, as well as ways in which they could be improved in future.

Prospective Challenges for Patenting and Clinical Trials of Anticancer Compounds from Natural Products: Coherent Review

Cancer is a leading cause of morbidity and mortality worldwide. Each year, millions of people worldwide are diagnosed with cancer, and more than half of them die. Various conventional therapies for cancer, including chemotherapy and radiotherapy, have extreme side effects. Therefore, to minimize the global burden of lethal diseases like cancer, an effective and novel drug must be discovered. Its patent should be acquired to secure the novel medicament. The pharmacological potential of different natural products has made them popular in the healthcare and pharmaceutical industries. Various anticancer compounds are obtained from natural sources such as plants, microbes, and marine and terrestrial animals, including alkaloids, terpenoids, biophenols, enzymes, glycosides, etc. The term "natural products" is defined as the product of secondary or non-essential metabolic processes produced by living organisms (such as plants, invertebrates, and microorganisms). Although more precise definitions of NPs exist, they do not always meet consensus. Others define NPs as small molecules (excluding biomolecules) that emerge from the metabolic reaction. A handful of effective compounds are used currently from natural or analog moieties, and many more are in clinical studies. There is an excellent need for patenting molecules from natural products as the hit lead molecules are derived, isolated, and synthesized from natural products. However, these naturally occurring products may not be patentable under the law because they come from nature. This review highlights why natural products and compounds are hard to patent, under what patent law criteria we can patent these natural products and compounds, patent procedural guideline sources and why researchers prefer publication rather than a patent. Here, various patent scenarios of natural products and compounds for cancer have been given.

TaxiBGC: a Taxonomy-Guided Approach for Profiling Experimentally Characterized Microbial Biosynthetic Gene Clusters and Secondary Metabolite Production Potential in Metagenomes

Biosynthetic gene clusters (BGCs) in microbial genomes encode bioactive secondary metabolites (SMs), which can play important roles in microbe-microbe and host-microbe interactions. Given the biological significance of SMs and the current profound interest in the metabolic functions of microbiomes, the unbiased identification of BGCs from high-throughput metagenomic data could offer novel insights into the complex chemical ecology of microbial communities. Currently available tools for predicting BGCs from shotgun metagenomes have several limitations, including the need for computationally demanding read assembly, predicting a narrow breadth of BGC classes, and not providing the SM product. To overcome these limitations, we developed taxonomy-guided identification of biosynthetic gene clusters (TaxiBGC), a command-line tool for predicting experimentally characterized BGCs (and inferring their known SMs) in metagenomes by first pinpointing the microbial species likely to harbor them. We benchmarked TaxiBGC on various simulated metagenomes, showing that our taxonomy-guided approach could predict BGCs with much-improved performance (mean F₁ score, 0.56; mean PPV score, 0.80) compared with directly identifying BGCs by mapping sequencing reads onto the BGC genes (mean F₁ score, 0.49; mean PPV score, 0.41). Next, by applying TaxiBGC on 2,650 metagenomes from the Human Microbiome Project and various case-control gut microbiome studies, we were able to associate BGCs (and their SMs) with different human body sites and with multiple diseases, including Crohn's disease and liver cirrhosis. In all, TaxiBGC provides an in silico platform to predict experimentally characterized BGCs and their SM production potential in metagenomic data while demonstrating important advantages over existing techniques. IMPORTANCE Currently available bioinformatics tools to identify BGCs from metagenomic sequencing data are limited in their predictive capability or ease of use to even computationally oriented researchers. We present an automated computational pipeline called TaxiBGC, which predicts experimentally characterized BGCs (and infers their known SMs) in shotgun metagenomes by first considering the microbial species source. Through rigorous benchmarking techniques on simulated metagenomes, we show that TaxiBGC provides a significant advantage over existing methods. When demonstrating TaxiBGC on thousands of human microbiome samples, we associate BGCs encoding bacteriocins with different human body sites and diseases, thereby elucidating a possible novel role of this antibiotic class in maintaining the stability of microbial ecosystems throughout the human body. Furthermore, we report for the first time gut microbial BGC associations shared among multiple pathologies. Ultimately, we expect our tool to facilitate future investigations into the chemical ecology of microbial communities across diverse niches and pathologies.

Artificial intelligence in microbial natural product drug discovery: current and emerging role

Covering: up to the end of 2022Microorganisms are exceptional sources of a wide array of unique natural products and play a significant role in drug discovery. During the golden era, several life-saving antibiotics and anticancer agents were isolated from microbes; moreover, they are still widely used. However, difficulties in the isolation methods and repeated discoveries of the same molecules have caused a setback in the past. Artificial intelligence (AI) has had a profound impact on various research fields, and its application allows the effective performance of data analyses and predictions. With the advances in omics, it is possible to obtain a wealth of information for the identification, isolation, and target prediction of secondary metabolites. In this review, we discuss drug discovery based on natural products from microorganisms with the help of AI and machine learning.