人工智能结合拉曼光谱技术快速无损检测肠道微生物
傅钰
时长:18:39 分会场:2019中国肠道大会 - 新技术大会
对肠道微生物,尤其是未培养肠道微生物,的检测和鉴定是肠道微生物研究中的重要环节。拉曼光谱作为化学分子的“指纹图谱”,可以对生物样品在溶液中进行无损、非标记、非接触的原位检测分析,具有其他方法所不具备的技术优势。拉曼技术能够在微生物单细胞水平,无损、非标记地实时监测细胞内核酸、蛋白质、脂类和代谢产物等的组成和变化,获得细胞全部的拉曼图谱信息,即“拉曼组”信息。一个细胞的“拉曼组”包含细胞内全部分子组成和物理状态等信息,其承载的多维信息对应于细胞内的转录组、蛋白组以及代谢组的信息,因此拉曼组可以依据其承载的指纹信息对微生物单细胞进行生理生化的精确表征。由于拉曼组承载了海量的信息,因此我们采用机器学习人工智能技术,让人工智能对大量的拉曼组数据进行学习和分析,实现准确的表征和分类鉴定。在前期工作中,我们搭建了适用于微生物的拉曼光谱智能识别鉴定的机器学习系统,结果表明利用该技术对微生物的分类正确率达到95.16%,并对同一种微生物是否耐药性的预测正确率达到83%以上。
傅钰
中科院微生物所
研究内容:以酿酒酵母(Saccharomyces cerevisiae)为模式生物,结合单分子(Single-molecule)技术,研究真核生物DNA复制,DNA损伤修复及衰老的分子机制,旨在回答生命现象的最基本问题,并为癌症等严重危害人类健康的疾病治疗提供理论依据。主要研究方向1)真核生物DNA复制中复制起始,延伸和终止的具体生化过程以及相应的重要调节因子的作用原理;2)DNA损伤的修复途径与相应的调控机制,着重于DNA复制依赖型DNA损伤修复;3)衰老的生物化学机制和表观遗传机理。
了解更多
更多视频
肠道菌群成像新工具的开发与应用
化学工具的发展可以为肠道菌群研究提供新的研究角度和工具,报告中将介 绍我们近期利用基于 D-型氨基酸的代谢探针,采用序贯标记的方式(sequential tagging with D-amino acid-based metabolic probes, STAMP)实现了粪菌移植过程中植入菌的成像追踪与在受体小鼠体内存活情况的评估,解决了粪菌移植中潜在的发挥治疗效用的细菌种类鉴别这一难题,为深入理解这一极为复杂但又潜力巨大的微生物疗法提供了研究工具, 也为这一疗法的步骤优化与标准化提供了研究方法。此外,还将介绍我们最近在体内菌群实时观察成像方面的工作进展。通过使用带有炔基的 D-型氨基酸探针,在体内代谢标记供体肠道菌群之后,再配合使用带有叠氮基团的红外 II 区荧光染料在体外进行点击化学反应,将荧光基团偶联到菌群表面。将这一菌群植入受体小鼠之后,目前已可以实现小鼠体内菌群的活体实时成像观察,并将成像方法拓展到了各类常见的致病菌与共生菌菌株的在体成像观察中。
王炜 时长:21:15
Nanopore Long Reads: Farewell to Blind Spots in Microbiome Research
吴昕 时长:27:57
USING PROXIMITY TO FIX ASSEMBLY
Background and Aims: The loss of long-range sequence contiguity in the process of NGS sequencing is an obstacle to understanding the structure and function of genomes and metagenomes. This obstacle negatively affects both clinical research efforts and virtually all microbiome-centric projects as much genetic information cannot be reconstructed from complex mixed microbial communities without culturing the constituent microbes. Methods and Results: The chromosome conformation capture method, Hi-C, restores chromosome-scale contiguity to large genome assemblies and enables the deconvolution of numerous genomes from mixed samples such as complex microbial communities. Hi-C captures genomic proximity interactions through in vivo crosslinking, followed by proximity-ligation and sequencing. Since the crosslinks occur inside intact cells, any two loci that interact by Hi-C must have originated in the same cell, and this data can be used to deconvolute high quality genomes directly from mixed populations. We have developed a metagenomic discovery platform that exploits Hi-C proximity-ligation data and have applied it to a number of diverse sample types. Our ProxiMeta Hi-C extracts large numbers of genomes directly from microbiome samples without culturing, which can also associate plasmids and phage with hosts and separate strains without culturing. Conclusions: We will discuss the application of Hi-C data to a number of diverse microbiome samples. For example, we have been able to discover hundreds of novel strains and species from fecal and infected samples, as well as identify novel plasmid/phage host interactions.We have been able to identify the host for a number of AMR genes within novel species, underscoring the value of proximity-ligation data to microbiome research.
Ivan Liachko 时长:20:08
评论