A new Artificial Intelligence-based approach to Antibiotic Discovery

自从发现青霉素以来，抗生素已成为现代医学的基础。然而，私营部门缺乏新抗生素的生产，加上抗生素耐药性不受控制地增加，是一个重大的全球公共卫生问题。预计到2050年，每年因耐药感染而死亡的人数将达到1 000万人。目前，大部分生物医学研究和治疗学的发展只集中在一小部分靶标上，而大多数先导药物化合物已经显示出相对短期的有效性。由于抗菌素耐药性水平的提高，源自食源性病原体的感染正变得越来越难以治疗。快速集约化耕作方式促进了广谱抗菌素的滥用，为抗生素耐药菌（ARB）和抗生素耐药基因（ARGs）提供了理想的选择压力。因此，人类食物链中动物源耐抗生素细菌的存在是一个重大的全球公共卫生问题，有几项研究报告了食用动物和产品被耐抗生素菌株定植和/或感染和污染，如耐甲氧西林金黄色葡萄球菌（MRSA）、耐抗生素弯曲杆菌和产生广谱β -内酰胺酶（ESBL）的肠杆菌科（即沙门氏菌、大肠杆菌）。霍乱是一种急性腹泻感染，由摄入受霍乱弧菌污染的食物或水引起。在世界范围内，估计有13亿人面临风险，每年大约发生130万至400万例病例，造成21 000至143 000人死亡。此外，对于这种细菌，不加选择地使用广谱抗生素会造成额外的威胁，其表现为病原体群体中抗菌素耐药性（AMR）谱的出现和扩散。该项目的目的是开发一种基于人工智能（AI）的方法，以扩大发现：(i)更广泛的治疗靶点和（ii）单个先导药物分子与耐药细菌中已确定的靶点具有高结合亲和力。研究计划：WP 1：下一代测序（NGS）生物信息学分析、数据挖掘和基于机器学习的统计建模将用于扫描不同食源性病原体（霍乱弧菌、大肠杆菌、沙门氏菌、金黄色葡萄球菌、肠球菌、弯曲杆菌）的基因组，以发现与AMR相关的新基因（新的治疗靶点）。博士项目将依赖并建立在最近获得的(2)GCRF和(2)Dottorini innovatuk - china资助的数据上。对于所有这些分离株，已经进行了传统的基于培养的筛选，抗生素敏感性测试（AST）和全基因组DNA测序，以便可以应用机器学习将基因型（AMR基因）与表型（AMR谱）关联起来。WP 2：鉴定的amr相关基因将在诺丁汉的国家生物膜创新中心（NBIC）实验室使用DNA重组技术和微生物学进行验证。WP3：具有明确AMR功能的基因将进行3D结构建模，并通过使用深度神经网络和ChEMBL， DrugTargetCommons, DrugBank， Broad Institute Drug Repurposing Hub数据库，我们将通过识别新药来扩大我们的抗生素库。WP4。药物效率以及新鉴定的药物将在NBIC上通过细菌抑制生长试验进行验证。

Since the discovery of penicillin, antibiotics have become the foundation of modern medicine. However, the lack of production of new antibiotics in the private sector together with an uncontrolled increase in antibiotic resistance represents a major global public health issue. It is projected that deaths attributable to resistant infections will reach 10 million per year by 2050. Nowadays, a large proportion of biomedical research and the development of therapeutics focuses only on a small fraction of targets for which most lead drug compounds have shown relatively short-term effectiveness.Infections originating from foodborne pathogens are becoming more difficult to treat due to increasing levels of antimicrobial resistance. Rapid and intensive farming practices promote indiscriminate use of broad-spectrum antimicrobials, providing ideal selection pressure for antibiotic-resistant bacteria (ARB) and antibiotic resistance genes (ARGs). The presence of antibiotic-resistant bacteria of animal origin in the human food chain is thus a major global public health issue, with several studies having reported food animals and products being colonized and/or infected and contaminated by antibiotic-resistant strains, such as methicillin-resistant Staphylococcus aureus (MRSA), antibiotic-resistant Campylobacter spp, and extended spectrum-beta-lactamase (ESBL) producing-Enterobacteriaceae (viz. Salmonella spp., Escherichia coli). Cholera is an acute diarrheal infection caused by ingestion of food or water contaminated with the bacterium Vibrio cholerae. Worldwide, 1.3 billion people are estimated to be at risk and approximately 1.3 to 4 million cases occur annually with 21,000 to 143,000 resulting in death. Also, for this bacterium, the indiscriminate use of wide-spectrum antibiotics creates an additional threat represented by the appearance and diffusion of antimicrobial resistance (AMR) profiles in the pathogen population.The aim of this project is to develop an Artificial Intelligence (AI)-based approach to broaden the possibility to discover: (i) a wider range of therapeutic targets and (ii) individual lead drug molecules showing high binding affinity to the identified targets in the resistant bacteria.Research Plan: WP 1: Next-generation sequencing (NGS) bioinformatics analysis, data mining, and statistical modelling powered by machine learning will be used to scan the genomes of different foodborne pathogens (V. cholerae, E. coli, Salmonella, S. aureus, Enterococcus, Campylobacter) to find new genes (new therapeutic targets) associated to AMR. The PhD project will rely and build on data collected on recently awarded (2) GCRF and (2) InnovateUK-China grants by Dottorini. For all these isolates conventional culture-based screening, antibiotic susceptibility tests (AST), and whole-genome DNA sequencing have been carried out so that machine learning can be applied to correlate the genotype (AMR genes) to the phenotype (AMR profiles). WP 2: The identified AMR-associated genes will be validated at the National Biofilms Innovation Centre (NBIC) labs in Nottingham using DNA recombinant techniques and microbiology. WP3: Genes with a clear AMR function will undergo 3D structure modelling and by using deep neural network and ChEMBL, DrugTargetCommons, DrugBank, Broad Institute Drug Repurposing Hub databanks we will expand our antibiotic arsenal by identifying new drugs. WP4. Drug efficiency together with the newly identified drugs will be validated at NBIC using bacteria inhibition growth assays.