Translational Informatics for Natural Products as Antidepressant Agents

Depression, a neurological disorder, is a universally common and debilitating illness where social and economic issues could also become one of its etiologic factors. From a global perspective, it is the fourth leading cause of long-term disability in human beings. For centuries, natural products have proven their true potential to combat various diseases and disorders, including depression and its associated ailments. Translational informatics applies informatics models at molecular, imaging, individual, and population levels to promote the translation of basic research to clinical applications. The present review summarizes natural-antidepressant-based translational informatics studies and addresses challenges and opportunities for future research in the field.

These natural products are isolated from various sources that include animals, fungi, plants, and microbes (Gunatilaka, 2006;Cragg and Newman, 2013;Genilloud, 2014;Atanasov et al., 2015;Katz and Baltz, 2016;Newman and Cragg, 2016;Futamura et al., 2017;Ikeda, 2017;Baltz, 2019;Singla et al., 2020;Apoorva et al., 2021;Bansal et al., 2021;Swarnkar et al., 2021). Unlike the chemical commercial drugs, laboratory synthesis of natural products with a heterogeneous composition and unique structures is an arduous task, albeit, enhanced stability and efficacy are reported for them. Moreover, there is often no evidence of microbial resistance, major toxic or side-effects for natural products-based therapeutics, and formulations (Moore et al., 2017). Essentially, these natural products include plant metabolic extracts and secondary metabolites extracted from different medicinal plants. Together, these are reported to demonstrate neuroprotective effects and reduce the risk of neurodegenerative diseases that strengthen their candidature as potential neuroprotective agents (Farahani et al., 2015;Fajemiroye et al., 2016;Martins, 2018;Bhandari et al., 2019). In light of the above facts, ongoing research efforts for depression management are focused on developing natural products-based anti-depressants.
In the recent past, tremendous progress in the interdisciplinary fields of medicinal science and computeraided high throughput screening of therapeutic targets has accelerated the ominous process of drug discovery (Singh et al., 2009;Igoli et al., 2014;Singla, 2014;Khan et al., 2020). For this, various therapeutic target databases, models, and tools have been constructed that are essential for the management of vast clinical data Liu et al., 2020;Yan et al., 2020;Chen et al., 2021;He et al., 2021). Together, these pave the way to the modern era of drug discovery and "big data," with a foundation of five pillars," namelyValue," "Variety," "Velocity," "Veracity," and "Volume" (Dhingra and Kumar, 2007). The preclinical and clinical data collected from patients suffering from depression along with the data procured from healthy individuals make up the big depression data. This big data can be utilized for future approaches to data-driven medicine for depression. Figure 1 illustrates the 5 V s of big data characterization for depression. Computer-Aided Drug Discovery (CADD) based on the "big data," surpasses the traditional methods of drug discovery that rely on ligand and structure-based drug designing tools (Sairam et al., 2002;Hazra et al., 2012;Joon et al., 2021). Howbeit, this demands a standardized multi-tiered data integration with robust algorithms for mining, structuring, and analysis of the accumulated clinical data Miao et al., 2021). Also, this is crucial for assessing the precision of drug-target interactions generated by molecular simulation studies. Further, these contemporary drug discovery databases are expected to have ample data storage capacity with proper sources to ascertain the actual data density for the discovery of novel drugs and their targets. Moreover, the appropriate utilization of these modernized computational approaches is crucial for novel drug discovery.
Even though much research has been done on natural product-based therapeutics that strengthen their credibility in the treatment of myriad ailments, the translation of data obtained from basic research to clinical application is still a challenge to the scientific community Singla et al., 2021a;Singla et al., 2021b;Singla et al., 2021c;Madaan et al., 2021). With a focus on the antidepressant application of these natural products, here, we propose a paradigm of the informaticsdriven research model and summarize the application of informatics models at different levels for augmenting translational research. Since the studies on the application of data models, Bioinformatics, imaging informatics, medical informatics, and health informatics in natural products-based drug discovery is scarce, this review will illuminate developmental insights for future novel translational informatics-based research directions in the field.  Frontiers in Cell and Developmental Biology | www.frontiersin.org January 2022 | Volume 9 | Article 738838 2 Depression pathogenesis involves multiple complex molecular mechanisms ( Figure 2). To date, two pathways, including mitogen-activated protein kinase signaling (MAPK) and cyclic adenosine phosphate signaling (cAMP), have been widely reported to be associated with depression development, which has attracted close attention in antidepressant research (Duman and Voleti, 2012).
The Ras-MAPK signaling pathway includes two key kinases, namely extracellular signal-regulated kinase (ERK) and MAP/ ERK kinase (MEK). ERK is typically activated by the phosphorylation of various growth factors. The activated ERK will then enter the nucleus and regulate a bunch of transcription factors, promoting specific genes' transcription and expression. ERK signaling is regulated by protein kinase A (PKA) and protein kinase C (PKC), the activators of which can activate the ERK 1/2 in the hippocampus (Roberson et al., 1999). Accumulating pieces of evidence have demonstrated that the decreased activity of ERK signaling is one of the contributors to depression. Data from studies on depressive suicide victims showed reduced activity of PKA, PKC, and adenylate cyclase (AC) (Dwivedi et al., 2004). Duric et al. sequenced the hippocampus tissues from major depressive disorder (MDD) patients and noticed a significantly elevated expression of MKP1, which is a suppressor for ERK and MEK (Duric et al., 2010). Moreover, Dwivedi et al. revealed the decreased catalytic and phosphorylation activity of ERK 1/2 via direct examination of the expression level of ERK 1/2 in the brain tissues from the suicide victims suffering from depression (Dwivedi et al., 2001). They also noticed the downregulation of Ras, which is an upstream regulator of ERK, and the interaction between Ras and MEK1 is limited as well (Dwivedi et al., 2009). Besides, MAPK suppressors like protein tyrosine phosphatase receptor type R (PTPRR) are also engaged in the ERK-involved depression mechanisms. Overexpression of PTPRR in mice led to their susceptibility to depression (Li et al., 2016). Taken together, ERK signaling plays a crucial role in enhancing neuron plasticity and promoting the release of neural growth factors. The general mechanisms of ERK signaling in depression development are due to its inactivation or suppression. Studies from a system level, however, should be addressed to further explore the landscape of ERK-engaged depression pathogenesis.
The cAMP signaling, or cAMP/AC/PKA signaling, is heavily involved in emotion regulation. A general route of this pathway starts from the activation of AC, which will result in the catalysis of ATP to cAMP, followed by the activation of PKA. The activated PKA will then phosphorylate the cAMP response element-binding protein (CREB) and finally regulate the gene transcription. Studies have noticed that the promotion of cAMP signaling may have antidepressant effects. For instance, Data from Jiang et al. showed that the immobility time of rats was significantly reduced by a PKA agonist named 8-BR-camp (Jang et al., 2018). Nico et al. used a cAMP analog that inhibits PKA but not cAMP and revealed that besides the activation of PKA, the elevation of cAMP can also promote the activity of this signaling pathway and enhance the antidepressant effects (Liebenberg et al., 2011). In addition, Wang et al. also showed similar results. By injection of Raleigh Alvin (rolipram), which is an inhibitor for phosphoric acid lipase 2-4 (PDE4) that can prevent the hydrolysis of cAMP, the researchers found the increased level of cAMP and phosphorylated CREB, and reduce of immobile time of the mice (Wang et al., 2019). Notably, the cAMP and MAPK signaling partially share the same proteins in depression development. There may be interactions between these two pathways, the relationship of which should be paid more attention to, especially for those antidepressant relevant research.

NATURAL PRODUCTS AGAINST DEPRESSION
The traditional system of medicine is the bedrock for several commercial drugs for depression and is based on natural products from various sources (Baird-Lambert et al., 1982;England, 1998;Hu et al., 2002;Hedner et al., 2006;Kochanowska et al., 2008;Zhao et al., 2016;Kochanowska-Karamyan et al., 2020;Singla, 2021). There is substantial evidence on the antidepressant activity of metabolic extracts and metabolites isolated from various medicinal plants (Farahani et al., 2015). The metabolic extracts are derived from distinct plant parts, such as leaves, flowers, roots, fruits (powdered or unripe), stem bark, bulb (powdered), whole plant, seed, petal, stigma, rhizome, hypocotyls, and etc. Reportedly, the plant secondary metabolites with antidepression activity belong to different classes of phytocompounds that mainly comprise alkaloids, flavonoids, furocoumarins, glycosides, polyphenols, saponins, triterpenoids, and xanthones (Farahani et al., 2015). Together, these execute the anti-depression activity or neuroprotective effects through different mechanisms that target neurological signaling pathways or molecules responsible for depressive disorders (Farahani et al., 2015). The antidepressant properties of some important natural products are discussed below.

Natural Products From Various Other Sources for Depression Management
Karamyan et al. demonstrated in vivo antidepressant activity for veranamine isolated from the marine sponge (Verongula rigida). Owing to its selective affinity towards sigma-1 and 5HT2B receptors, it could serve as a novel antidepressant drug candidate (Kochanowska-Karamyan et al., 2020). Similarly, barettin, 8,9 dihydrobarettin, gelliusines A and B, and sigmaconotoxin have been reported to possess selective affinity towards the serotonin receptors (England, 1998;Hedner et al., 2006). Further, Lambert et al. reported the MAO inhibitory activity for methylaplysinopsin isolated from Aplysinopsis reticulate (a sponge) (Baird-Lambert et al., 1982). The antagonist binding displacement activity at 5HT 2A and 5HT 2C receptors have been reported for the compounds isolated from Smenospongia aurea (a sponge) (Hu et al., 2002). There are some other marine natural products reported to possess promising antidepressant activity in vivo (Kochanowska et al., 2008). For example, potential in vivo antidepressant activity was reported for the total sterols and β-sitosterol isolated from Sargassum horneri (a brown seaweed). There occurred a significant increase in NE, 5-HT, and 5-HIAA (5-hydroxyindoleacetic acid) neurotransmitters (Zhao et al., 2016).

DATABASES COMPRISING NATURAL PRODUCTS-SHARING OF DATA AND REFERENCES
With the breakthrough technologies, such as omics and informatics, biomedical research has made significant strides (Canuel et al., 2014;McGuire et al., 2020). Since the collected biomedical information is huge, data management with highly organized databases is enormously important (Harel et al., 2011). Besides, such well-structured databases must be amenable to sharing and integration of the standardized and annotated stored data (Misra et al., 2019). Natural products comprise a spectrum of potential therapeutic compounds for myriad diseases. Owing to the spectacular diversity in natural products, the development of natural-products databases is essential (Mehbub et al., 2014;Atanasov et al., 2015;Cheesman et al., 2017). These databases, in turn, will broaden FIGURE 5 | Illustration of the molecular mechanisms mediated by the bioactives for depression management.
Frontiers in Cell and Developmental Biology | www.frontiersin.org January 2022 | Volume 9 | Article 738838 7 our horizon on the mechanistic insights of natural products or compounds on a particular disease target and reveal crucial clinical details for 'precision medicine'. One such example is COCONUT, an acronym for MongoDB COlleCtion of Open Natural prodUcTs (https://coconut.naturalproducts.net) that comprise freely accessible natural products databases, albeit partially (Sorokina et al., 2021).

Traditional Medicine Databases
Based on the geographical location, the traditional medicinal system has various distinct branches with shared attributes. To name a few, these conventional medicinal systems include traditional Indian medicine, traditional Chinese medicine, and traditional Islamic medicine (Pan et al., 2014;Yuan et al., 2016). From a holistic perspective, the development of robust databases and knowledge bases is imperative for a systematic sharing and annotation of these traditional medicinal systems that encompass enormous information on natural products. This indeed is indispensable for an integrated evaluation and screening of natural products (Ikram et al., 2015). Over the years, endeavours have been undertaken to build compendious natural products databases that accommodate essential data on the natural products, their potential targets, and genetic interactions. HERB (http://herb.ac.cn) and SymMap (https:// www.symmap.org/) are exemplary traditional Chinese medicine databases. HERB contains elementary information on herbs with their putative targets and genetic interaction mechanisms while SymMap is useful in mapping the disease symptoms disseminating appropriate prescriptions Fang et al., 2021). Others in the category are the curated Indian Medical Plants, Phytochemistry And Therapeutics (IMPPAT) (https://cb.imsc.res.in/imppat), and Universal Natural Product Resource (UNaProd) (http://jafarilab.com/ unaprod) databases that provide useful information on the nomenclature and medicinal applications of Indian and Iranian herbs, respectively (Mohanraj et al., 2018;Naghizadeh et al., 2020). Noteworthy, data on natural products contained within these traditional medicine databases may provide us with valuable conventional clinical and therapeutic anti-depression prescriptions. Howbeit, a real-time integrated data analysis is allimportant for gaining newer clinical insights to the global scientific community dedicated to depression research.

TRANSLATIONAL INFORMATICS FOR INVESTIGATION OF POTENTIAL NATURAL PRODUCTS AS ANTI-DEPRESSANTS-DATA INTEGRATION AND MODELING
Computational tools based on multi-scale modeling (MSM) are explicitly efficient, robust, and dynamic in integrating data, testing hypotheses, and comprehensively illuminating the pathophysiological mechanisms underlying depressionrelated neurological disorders (Ramirez-Mahaluf et al., 2015;Shen et al., 2019). Collectively, these expedite diagnosis and therapy together with antidepressant target identification for drug development (Ramirez-Mahaluf et al., 2015). These putative antidepressant molecules can be explored to ascertain their candidature as a proficient therapeutic target or biomarker for depression disorders by employing amalgamated and synchronized network-based strategies (Zhang T.-T. et al., 2018;Wu et al., 2018). Over the past few years, deep learning has revolutionized the traditional target screening, which embarks on a new age of drug discovery (Schneider, 2017;Kraus, 2019;Kiriiri et al., 2020;Schaduangrat et al., 2020;Gupta et al., 2021;Paul et al., 2021;Singh et al., 2021). For instance, TripletRes and AlphaFold are globally acclaimed contemporary twodimensional (2D) protein structure prediction tools of deep learning, which have spectacularly boosted the efficacy of classical drug discovery strategies (Kryshtafovych et al., 2019;Li et al., 2019;Senior et al., 2019). A consolidated yet thoughtful utilization of these modern computational tools will be decisive in exploring naturally occurring molecules for depression management and prognosis (Truax and Romo, 2020;Woo and Shenvi, 2021).

Computational Models for the Synthesis of Natural Products
In general, the naturally occurring biomolecules are either isolated from microbial fauna or medicinal plants, which is an expensive and labour extensive long rigmarole (Sairam et al., 2002;Sarker and Nahar, 2012;Bucar et al., 2013;Zhang Q.-W. et al., 2018). These cumbersome isolation procedures are oftentimes undermined by inevitable constraints, such as the seasonal variations of plant growth, variations in microbial growth conditions (in the case of microbial biomolecules), the efficacy of the purification procedures employed, and low yields (Sairam et al., 2002;Sarker and Nahar, 2012;Bucar et al., 2013;Zhang Q.-W. et al., 2018). A combination of highly efficient isolation and purification techniques is, therefore, indispensable to obtain these natural products at reasonable yields. Even though there exist a few fully autonomous computational algorithms and tools, these are tarnished by major pitfalls. For instance, these are capable of accomplishing solo commands at a given time and are usually confined to comparatively simple molecular drug targets (Song et al., 2009;Prachayasittikul et al., 2015;Bharatam, 2021). A highly acclaimed synthesis route design tool for complex natural products is Chematica (Mikulak- Klucznik et al., 2020). With the innovation of Chematica, autonomous-computer-aided synthesis pathways are designed swiftly for myriad commercially important natural products and biologically active compounds of medicinal value. In contrast to the previous synthesis pathways, it requires fewer steps with incredible synthetic efficiency and costeffectiveness evident from its laboratory performance (Klucznik et al., 2018).
It is worth discussing here, that the classical retro-synthesis technique for even simple organic molecules relied on the recursive or repetitive transformation into still smaller entities. This cumbersome task has been hastened by the advent of computer-based retro-synthesis. Quite dismally, these are in their infancy with disappointingly sluggish performance and quality-compromised outputs. Recently, these limitations have been overcome with the introduction of symbolic artificial intelligence (AI) and Monte Carlo tree search-guided revelation of retro-synthesis routes for diverse organic molecules (Segler et al., 2018). This state-of-the-art computer-aided retro-synthesis tool essentially consists of deep neural networks, viz. expansion policy, and filter networks integrated Monte Carlo tree search to allow a guided search with the prior selection of the propitious steps for synthesis route. Further, these deep learning tools are exceptionally swift in their output for diverse molecules over manually designed conventional heuristic methods that rely on extracted rules for synthesis route search. Also, these are well-trained, with an inbuilt knowledge on nearly all the reported organic reactions for myriad molecules as ascertained by a double-blind analysis (Segler et al., 2018).
This spectacular breakthrough in the field of computer-aided retro-synthesis might assist researchers in devising novel techniques to deduce feasible approaches for the optimal synthesis of molecular targets. Also, this does not require any prior knowledge or expertise regarding the existing strategies.
Frontiers in Cell and Developmental Biology | www.frontiersin.org January 2022 | Volume 9 | Article 738838 Altogether, these contemporary computational tools, perhaps strategize the multi-step complex syntheses route designs for natural products, which are otherwise quite laborious and inefficacious.
Computational Models for Natural Products-Based "Precision Medicine" The network-based approaches play a cardinal role in numerous scientific fields (Chandran et al., 2017;Guo et al., 2020). Amongst the various crucial applications is in the domain of biomedical sciences, where it assists in the evaluation of diverse systemic molecular interactions (Sonawane et al., 2019;Sheik Amamuddy et al., 2020;Wang et al., 2020). To perform investigations, such as assessing the effects of dysfunctional molecules in the system as a whole, hitherto biological networks were relied upon (Furlong, 2013;Somvanshi and Venkatesh, 2013;Altaf-Ul-Amin et al., 2014;Charitou et al., 2016;Hu et al., 2016;Caldera et al., 2017;Faeder et al., 2020;Silverman et al., 2020;Wang et al., 2020). The discovery of biomarkers together with a screening of putative drug molecules for complex diseases, including depression-related disorders can be accomplished using more utilitarian network-based applications. These include, but are not limited to co-expression, gene-gene, and protein-protein interaction network-based strategies (Khanin et al., 2011;Vella et al., 2017;Myers et al., 2019;Sun et al., 2019;Ovens et al., 2021).
With the innovation of genomics or genomic sequencing (DNA/RNA), the process of novel drug discovery is tremendously accelerated (Xia, 2017;Suwinski et al., 2019;McGuire et al., 2020). Further, genome sequencing was pioneering in introducing the concept of "Drug repurposing" for medically-approved drugs, which was a turning point in treating diverse ailments and a diminished economic burden for developing newer drugs for individual disease treatment (Emilien, 2000;Jarada et al., 2020;Nabirotchkin et al., 2020). Howbeit, their applicability in the therapeutic management of numerous disparate diseases was sceptical. This limitation is overcome by the contemporary and technologically advanced genomic techniques that sped the identification of specific disease-causing key genetic factors or anomalies in an individual, in particular (Dryja, 1997;Lander et al., 2001;Hasin et al., 2017;Horton and Lucassen, 2019). This "Precision medicine" furnishes a mechanistic insight into an individual patient's disease (customized disease module) and unveils principal disease contributing elements. Eventually, these disease mechanisms could be targeted with "precision" for high-end personalized treatment strategies (Dugger et al., 2017;König et al., 2017). In this direction, a drug repurposing, Genome-wide Positioning Systems network (GPSnet) algorithm that targets genomic sequence profile-derived disease modules from a single patient was developed. These genomic sequencing profiles allow protein-protein interaction mapping for human diseases that reveal the key molecular players in disease pathophysiology. This, in turn, is pivotal in strategizing and prioritizing the selection of repurposed drugs for an effective and customized treatment regimen. Further, disease modules based on the predictions from GPSnet (https://www.gpsnet. com.br/) could accurately predict responses and strategize usage for a reasonable number of approved chemotherapeutic drugs for approximately five thousand cancer patients on prior in silico investigations (transcriptomic profiling and exome sequencing) . As a proof of concept, ouabain (cardiac drug) demonstrated an antitumor potency via anti-HIF1α/LEO1 activity in vitro. This in silico tool could perform the dual role of specifically identifying a disease module and repurposing the approved drugs with precise indications for medical applications as observed in the case of cancer. These findings strengthen the candidature of GPSnet as a drug repurposing scaffold for constructing an effective therapeutic screen for various drugs, including identification of naturally occurring potential drug molecules, their synthesis, and precise administration in depression therapies .
There is accumulating evidence on the utility of network theory in evaluating the therapeutic potential of natural products in health management. For instance, association network-based novel techniques were developed for scrutinizing and discerning microbes that synthesize biomolecules from those that participate in the biological transformation of natural (or pharmaceutical) products within the human host. Specifically, these association networks rely on the concomitant probing of metabolomics and metagenomics data on diversified human microbial fauna. As a step further, the intended molecules were mapped to their respective clade and finally to the phylogenetic tree to identify the microbial species participating in their synthesis or biotransformation (Cao et al., 2019). As a futuristic approach, these might supersede the available time and cost-extensive characterization techniques, which are solely dependent on methods of microbial cultivation. Further, this limits their efficiency of species identification that partakes in the synthesis or the transformation of the vivid small molecular wealth found within the host system. The feasibility study of metagenomics and metabolomics association networks unveiled the corynomycolenic acid-producing microbial genes amid the human cystic fibrosis microbiome isolates. Additionally, these accurately delineated the associations of quinolone signals (Pseudomonas), phevalin, and tyrvalin to their respective clusters of biosynthetic genes (Cao et al., 2019).
In another study, the mechanisms that govern natural products and synthetic chemotherapeutic synergism were investigated. Accordingly, the information on compounds and their targets was retrieved from the public domain that aided in assessing the targetable space for respective natural products. In the context of the network, their evaluation accentuated the notion that these natural products exhibit groupings of targets in the family, which are disparate as well as share commonness with a synthetic chemotherapeutic. Conclusively, these rational pieces of evidence emphasize the chemotherapeutic efficiency of natural products for developing complementary and combinatorial novel chemotherapies with synthetic anticancer drugs (Chamberlin et al., 2019). Likewise, multi-potent natural chemotherapeutics were screened from Clerodendrum indicum and C. serratum using network pharmacology (Gogoi et al., 2017). By employing an integrative approach, the anticancer effects of a combination of drugs against various cancer targets were determined. Amongst the predicted natural anti-cancer compounds, apigenin 7-glucoside and hispidulin could bind efficiently to reasonable chemotherapeutic targets (seventeen). These findings are crucial milestones in the field of novel anticancer drug discovery (Gogoi et al., 2017).
Whilst shreds of evidence on network-based approaches for anti-depression natural product discovery are lacking, the above findings might serve as important milestones in bridging the gap in the identification of natural products and developing these as potent anti-depression therapeutics.

FUTURE PERSPECTIVES ON TRANSLATIONAL INFORMATICS FOR INVESTIGATION OF NATURAL PRODUCTS ANTI-DEPRESSANTS
With a global technological advancement in the healthcare sector, depression management strategies now include translational informatics, which has firmly integrated clinical data with basic research (Smith et al., 2007;Unützer and Park, 2012;Tenenbaum, 2016;Robinson, 2018;Kraus et al., 2019). Also, these aim to strive at "precision medicine" for depression care (Nierenberg, 2012;Menke, 2018;Serretti, 2018). Noteworthy, these substantially rely on multi-tiered databases comprising enormous yet, segregated data from various resources, such as clinical, environmental, lifestyle, and natural products data (Herland et al., 2014;Shameer et al., 2017;Seyhan and Carini, 2019). These, in turn, are essential for developing a well-trained AI system with ample inbuilt knowledge. It is contemplated that the AI system, so developed, can be exploited for performing dual tasks of exploring potential natural therapeutic candidates commenced by recommendation of precise drugs for depression management at first. Secondly, the real-time medical status of a single patient can be monitored based on the instantaneous physiological information together with an automated AI system alarm upon encountering health aberrations. Furthermore, healthcare counselling and related advice can be imparted to patients by day-to-day evaluation using cloud computing. Unfortunately, the AI system for depression management is still in its infancy struggling with issues of privacy and confidentiality of patient information amongst myriad other challenges (Grist et al., 2018;Graham et al., 2019;Hategan et al., 2019;Huckvale et al., 2019;Romano and Tatonetti, 2019;Tran et al., 2019;Bickman, 2020;Ke et al., 2020;Mennen et al., 2021). Figure 6 illustrates the futuristic translational informaticsbased model for depression management using natural antidepressants. Presently, the databases on natural products store data on the promising candidates for myriad diseases. With the increasing research data on depression-related therapeutics based on natural products and the complexity associated with the disease, it is important to integrate these databases with systematic analysis. These, in turn, are expected to disseminate the references and knowledge for well-trained AI systems. It is, therefore, contemplated that the AI systems can be exploited for performing two important tasks. Firstly, these can be utilized for natural product screening and for endorsing promising candidates for depression-targeted pre-clinical and clinical trials. Next, the effective anti-depression natural products, as revealed by these studies, are included in the treatment regime of the patients diagnosed with depression. Depending upon their effectiveness, these can be administered either as a stand-alone or in combination with commercially approved drugs. Secondly, AI systems can be used for surveillance as well as monitoring the health status of the patients. Accordingly, patients' physiological data are acquired in realtime through wearables and cloud platform-assisted technology. This caters to the customized needs of the patients by providing them suggestions for their self-care, and their health status is also reported simultaneously. Although this systemic translational informaticsbased model for depression management using natural antidepressants is promising and attractive, various issues are yet to be addressed to implement this futuristic approach.

Databases and Knowledge Bases for Specific Antidepressant Natural Products
Unfortunately, the available databases for natural products contain scarce anti-depression therapeutic candidates (Varteresian and Lavretsky, 2014;Sorokina et al., 2021). As the scientific intrigue and general concern for depression management have gained momentum, it is imperative to develop anti-depression natural therapeutic candidate databases. For futuristic computational therapeutic screening, these depression-specific databases are essential. In light of the disease complexity, these databases must mandatorily employ an integrated systematic analysis. Further, the construction of knowledge databases is suggested as they undertake the compilation of data obtained from distinct levels. These include, but are not limited to, potential therapeutic biomolecules with the databases of natural products, related biomolecules, and their putative targets within the host system, environmental factors, and various other depression-associated attributes. Eventually, a stratified biomedical landscape for depression management is sketched by these knowledge databases. These knowledge databases, in turn, aid in the construction of knowledge graphs, which are expert-populated data integration biomedical resources. Specifically, the biomedical entities (concepts) are illustrated as nodes while inter-entity associations or relationships are depicted as edges (Yu et al., 2017;Nicholson and Greene, 2020). These might be of immense significance in assisting various biomedical applications capable of comprehending novel clinical, genomic, and pharmaceutical details needed for treatment support decisions.

Systematic Modeling Based on AI Screening of Potential for Antidepressant Natural Products
Since depression is a neurological and debilitating disease with complex traits, which might have genetic roots as observed in the case of other psychiatric disorders such as cystic fibrosis and Huntington's chorea (Bowcock, 2010;Dunn et al., 2015;Shadrina et al., 2018). Owing to the complexity of contributing factors, the AI system largely reckons on molecular dynamic simulations-and modeling tools-based depression evolutionary analysis for evaluating the implications of natural products for patients with depression disorders (Romano and Tatonetti, 2019). Consequently, unveiling the principal elements and crucial molecular players associated with disease progression and deciphering effective natural products for its management becomes an onerous task. This generates mind-boggling yet critical questions concerning improvisations on the robustness of AI for screening of natural products and precision medicine . Quite possibly, these can be subdued by employing quality training data, selection, optimization, and validation algorithms, feature extraction as well as standard and validated techniques-based data collection.

Cross-Level Data Integration-dependent Precision Treatment for Depression
Depression-specific medicinal research encompasses complex clinical and molecular phenotypic data types with yet more complicated interconnections (Krishnan and Nestler, 2008;Clark et al., 2017). By and large, the ongoing research emphasizes certain aspects while neglecting other facets, which might be commensurably important in prompt diagnosis, therapy, and management of depressive disorders. Mapping these entwined clinical and molecular phenotypic linkages is cardinal for modeling in systems biology for any disease type (Schumann et al., 2014). For instance, there have been reports on the therapeutic efficacy of some natural products at the molecular level without any marked effect on the clinical phenotypes of patients (Liu et al., 2015;Yeung et al., 2018). This supports the notion that gathering a paired molecular and clinical data for a defined duration can be instrumental in simulating disease progression, depression, in this case, and generating a reliable model for the same. The Cancer Genome Atlas (TCGA) (https:// www.cancer.gov/) and the International Cancer Genome Consortium (ICGC) (https://daco.icgc.org/) are exemplary data integration programs that can be viewed as a landmark for constructing similar depression-disorder programs. In conclusion, Figure 6 represents a promising systemic model for depression management, however, the feasibility of this model relies on several challenges, including the scarce quantitative data on natural antidepressants.

AUTHOR CONTRIBUTIONS
RS, SJ, and LS collected the data, analyzed it, and wrote the manuscript. BS generated the idea and outline, fundraiser, and finally revised the manuscript.

FUNDING
This work was supported by the National Natural Science Foundation of China (32070671), the Covid-19 research projects of West China Hospital Sichuan University (Grant No. HX-2019-nCoV-057) as well as the regional innovation cooperation between Sichuan and Guangxi Provinces (2020YFQ0019).