FooDrugs database: A database with molecular and text information about food - drug interactions

FooDrugs database is a development done by the Computational Biology Group at IMDEA Food Institute (Madrid, Spain), in the context of the Food Nutrition Security Cloud (FNS-Cloud) project. Food Nutrition Security Cloud (FNS-Cloud) has received funding from the European Union's Horizon 2020 Research and Innovation programme (H2020-EU.3.2.2.3. – A sustainable and competitive agri-food industry) under Grant Agreement No. 863059 – www.fns-cloud.eu (See more details about FNS-Cloud below) FooDrugs stores information extracted from transcriptomics and text documents for foo-drug interactiosn and it is part of a demonstrator to be done in the FNS-Cloud project. The database was built using MySQL, an open source relational database management system. FooDrugs host information for a total of 161 transcriptomics GEO series with 585 conditions for food or bioactive compounds. Each condition is defined as a food/biocomponent per time point, per concentration, per cell line, primary culture or biopsy per study. FooDrugs includes information about a bipartite network with 510 nodes and their similarity scores (tau score; https://clue.io/connectopedia/connectivity_scores) related with possible drug interactions with drugs assayed in conectivity map (https://www.broadinstitute.org/connectivity-map-cmap). The information is stored in eight tables: Table “study” : This table contains basic information about study identifiers from GEO, pubmed or platform, study type, title and abstract Table “sample”: This table contains basic information about the different experiments in a study, like the identifier of the sample, treatment, origin type, time point or concentration. Table “misc_study”: This table contains additional information about different attributes of the study. Table “misc_sample”: This table contains additional information about different attributes of the sample. Table “cmap”: This table contains information about 70895 nodes, compromising drugs, foods or bioactives, overexpressed and knockdown genes (see section 3.4). The information includes cell line, compound and perturbation type. Table “cmap_foodrugs”: This table contains information about the tau score (see section 3.4) that relates food with drugs or genes and the node identifier in the FooDrugs network. Table “topTable”: This table contains information about 150 over and underexpressed genes from each GEO study condition, used to calculate the tau score (see section 3.4). The information stored is the logarithmic fold change, average expression, t-statistic, p-value, adjusted p-value and if the gene is up or downregulated. Table “nodes”: This table stores the information about the identification of the sample and the node in the bipartite network connecting the tables “sample”, “cmap_foodrugs” and “topTable”. In addition, FooDrugs database stores a total of 6422 food/drug interactions from 2849 text documents, obtained from three different sources: 2312 documents from PubMed, 285 from DrugBank, and 252 from drugs.com. These documents describe potential interactions between 1464 food/bioactive compounds and 3009 drugs. The information is stored in two tables: Table “texts”: This table contains all the documents with its identifiers where interactions have been identified with strategy described in section 4. Table “TM_interactions”: This table contains information about interaction identifiers, the food and drug entities, and the start and the end positions of the context for the interaction in the document. FNS-Cloud will overcome fragmentation problems by integrating existing FNS data, which is essential for high-end, pan-European FNS research, addressing FNS, diet, health, and consumer behaviours as well as on sustainable agriculture and the bio-economy. Current fragmented FNS resources not only result in knowledge gaps that inhibit public health and agricultural policy, and the food industry from developing effective solutions, making production sustainable and consumption healthier, but also do not enable exploitation of FNS knowledge for the benefit of European citizens. FNS-Cloud will, through three Demonstrators; Agri-Food, Nutrition & Lifestyle and NCDs & the Microbiome to facilitate: (1) Analyses of regional and country-specific differences in diet including nutrition, (epi)genetics, microbiota, consumer behaviours, culture and lifestyle and their effects on health (obesity, NCDs, ethnic and traditional foods), which are essential for public health and agri-food and health policies; (2) Improved understanding agricultural differences within Europe and what these means in terms of creating a sustainable, resilient food systems for healthy diets; and (3) Clear definitions of boundaries and how these affect the compositions of foods and consumer choices and, ultimately, personal and public health in the future. Long-term sustainability of the FNS-Cloud will be based on Services that have the capacity to link with new resources and enable cross-talk amongst them; access to FNS-Cloud data will be open access, underpinned by FAIR principles (findable, accessible, interoperable and re-useable). FNS-Cloud will work closely with the proposed Food, Nutrition and Health Research Infrastructure (FNHRI) as well as METROFOOD-RI and other existing ESFRI RIs (e.g. ELIXIR, ECRIN) in which several FNS-Cloud Beneficiaries are involved directly. (https://cordis.europa.eu/project/id/863059) ***** changes between version FooDrugs_v2 and FooDrugs_V3 (31st January 2023) are: Increased the amount of text documents by 85.675 from PubMed and ClinicalTrials.gov, and the amount of Text Mining interactions by 168.826. Increased the amount of transcriptomic studies by 32 GEO series. Removed all rows in table cmap_foodrugs representing interactions with values of tau=0 Removed 43 GEO series that after manually checking didn't correspond to food compounds. Added a new column to the table texts: citation to hold the citation of the text. Added these columns to the table study: contributor to contain the authors of the study, publication_date to store the date of publication of the study in GEO and pubmed_id to reference the publication associated with the study if any. Added a new column to topTable to hold the top 150 up-regulated and 150 down-regulated genes.

FooDrugs数据库由西班牙马德里IMDEA食品研究所（IMDEA Food Institute）计算生物学团队在食品营养安全云（Food Nutrition Security Cloud, FNS-Cloud）项目框架下开发。该项目已获得欧盟地平线2020（Horizon 2020）研究与创新计划（H2020-EU.3.2.2.3.——可持续且具竞争力的农业食品产业）的资助，资助协议编号为863059，项目官网为www.fns-cloud.eu（FNS-Cloud的更多细节详见下文）。

FooDrugs存储从转录组学与文本文档中提取的食品-药物相互作用信息，是FNS-Cloud项目待开发的演示系统的组成部分。该数据库基于开源关系型数据库管理系统MySQL构建。

FooDrugs共收录161组转录组学GEO（Gene Expression Omnibus）系列数据，涵盖585种食品或生物活性化合物处理条件。每种条件被定义为：每项研究中，对应特定时间点、浓度、细胞系、原代培养物或活检样本的食品/生物组分。

FooDrugs包含一个含510个节点的二分网络信息，以及与连接图谱（Connectivity Map, CMAP，https://www.broadinstitute.org/connectivity-map-cmap）中检测药物相关的潜在药物相互作用有关的相似性评分——tau评分（tau score，https://clue.io/connectopedia/connectivity_scores）。

该数据库的信息存储于8张数据表中：
1. "study表"：存储来自GEO、PubMed或平台的研究标识符、研究类型、标题及摘要等基础信息。
2. "sample表"：存储研究中不同实验的基础信息，如样本标识符、处理方式、来源类型、时间点或浓度。
3. "misc_study表"：存储研究各项附加属性的补充信息。
4. "misc_sample表"：存储样本各项附加属性的补充信息。
5. "cmap表"：存储70895个节点的相关信息，涵盖药物、食品或生物活性物质、过表达及敲低基因（详见3.4节），信息包括细胞系、化合物及扰动类型。
6. "cmap_foodrugs表"：存储关联食品与药物或基因的tau评分（详见3.4节），以及FooDrugs二分网络中的节点标识符。
7. "topTable表"：存储每项GEO研究条件下的150个上调及下调基因信息，用于计算tau评分（详见3.4节），存储内容包括对数倍变化值、平均表达量、t统计量、p值、校正后p值，以及基因的上调/下调状态。
8. "nodes表"：存储样本及二分网络节点的标识信息，用于连接"sample"、"cmap_foodrugs"与"topTable"三张表。

此外，FooDrugs数据库还存储了来自2849份文本文档的6422条食品/药物相互作用数据，这些文档来源于三个不同渠道：PubMed的2312份文档、DrugBank的285份文档，以及drugs.com的252份文档。这些文档描述了1464种食品/生物活性化合物与3009种药物之间的潜在相互作用。相关信息存储于两张数据表中：
1. "texts表"：存储所有已通过4节所述策略识别出相互作用的文档及其标识符。
2. "TM_interactions表"：存储相互作用标识符、食品与药物实体，以及文档中相互作用上下文的起始与终止位置信息。

FNS-Cloud将通过整合现有食品营养安全（FNS）数据解决数据碎片化问题，这对于开展高端泛欧洲FNS研究至关重要，研究范畴涵盖FNS、膳食、健康与消费者行为，以及可持续农业与生物经济。当前碎片化的FNS资源不仅造成知识缺口，阻碍公共卫生与农业政策制定、食品工业开发有效解决方案，无法实现生产可持续化与消费健康化，还无法利用FNS知识造福欧洲公民。

FNS-Cloud将通过三大演示项目——农业食品、营养与生活方式，以及非传染性疾病（NCDs）与微生物组——助力实现以下目标：
(1) 分析区域及国家层面的膳食差异，包括营养、（表观）遗传学、微生物组、消费者行为、文化与生活方式及其对健康的影响（肥胖、非传染性疾病、民族与传统食品），这对于公共卫生、农业食品与健康政策至关重要；
(2) 加深对欧洲内部农业差异的理解，以及这些差异对构建可持续、韧性的健康膳食食品系统的意义；
(3) 明确边界定义及其对食品成分、消费者选择的影响，最终影响未来个人与公共健康。

FNS-Cloud的长期可持续性将依托能够对接新资源并实现资源间互通的服务体系；其数据将采用开放获取模式，遵循FAIR原则（可发现、可访问、可互操作、可复用）。

FNS-Cloud将与拟议的食品、营养与健康研究基础设施（FNHRI）、METROFOOD-RI以及其他现有欧洲战略论坛研究基础设施（ESFRI RIs，如ELIXIR、ECRIN）紧密合作，多个FNS-Cloud项目受益方已直接参与这些基础设施。项目详情可参考：https://cordis.europa.eu/project/id/863059

***** FooDrugs_v2与FooDrugs_V3（2023年1月31日更新）的变更内容如下：
1. 从PubMed及ClinicalTrials.gov新增的文本文档数量增加85.675份，文本挖掘交互数据量增加168.826条；
2. 转录组学研究数量新增32组GEO系列数据；
3. 移除cmap_foodrugs表中所有tau值为0的相互作用条目；
4. 手动核查后移除43组与食品化合物无关的GEO系列数据；
5. 为texts表新增"citation"列，用于存储文本的引用信息；
6. 为study表新增三列："contributor"列存储研究作者，"publication_date"列存储研究在GEO及PubMed平台的发表日期，"pubmed_id"列关联研究对应的发表文献（若有）；
7. 为topTable表新增列，用于存储前150个上调基因与前150个下调基因。