DNA 3.0: Developing novel enzymes for DNA synthesis with deep learning and combinatorial genetics

项目总结/摘要 DNA合成在生物技术革命中发挥了关键作用。很容易获得 合成的DNA寡核苷酸和由它们组装的基因，对于 阐明和解锁生物功能，并使合成生物学的新领域， 可以创造新的细胞、酶、治疗剂、诊断剂和其他具有商业价值的试剂。 尽管有这种影响，DNA合成使用的是30多年前开发的化学策略， 是昂贵的并且限于长度为200个核苷酸或更少的分子。 下一代酶促DNA合成技术正在探索中， 独立DNA聚合酶（TIDP），用于将核苷酸受控地添加到生长的DNA中 搁浅。尽管最近已经报道了进展，但酶促DNA合成仍然受到以下限制： 可用的TIDP的低效率，特别是这些聚合酶的相对无能 以掺入3 '-封闭的核苷酸。 在这个第一阶段的小企业创新研究（SBIR）项目中，原始遗传学公司， 合成生物学公司与差异化组合遗传技术，和Denovium公司， 一家开创基因发现新方法的人工智能公司， 力开发新的和高效的TIDPs酶促DNA合成在体外。 Denovium将利用其基于机器学习算法的计算能力， 从专有和公共数据库中发现具有所需活性的新型TIDP。德诺维乌姆 还将进行专有的人工智能（AI）扫描，以确定所有功能的影响， 选择的TIDP上的可能突变。原始遗传学将合成并表达 收集序列，并在体外进行测试，以确定最活跃的酶。最好的 2种酶将使用原始遗传公司专有的函数发生器技术进行多样化 以及其他随机多样化方法。编码酶变体的基因群体将 用超高通量筛选来鉴定最具活性的酶。数据集 这项工作的结果将用于训练Denovium的序列预测算法，以加速 在第二阶段进一步改进酶。 这项工作是分离和开发新的酶的可行性研究， 酶促DNA合成，也用于创建酶优化工具的管道。的 在这项工作中发现的酶将直接用于酶促DNA合成应用， 并且可以被许可或出售给领先的DNA和基因制造商。

Project Summary/Abstract DNA synthesis has played a key role in the biotechnology revolution. The ready availability of synthetic DNA oligonucleotides and of genes assembled from them, has been invaluable for elucidating and unlocking biological function and enabling the new field of synthetic biology which can create novel cells, enzymes, therapeutics, diagnostics and other reagents of commercial value. Despite this impact, DNA synthesis uses chemical strategies developed over 30 years ago which are costly and limited to molecules of 200 nucleotides or less in length. Next-generation enzymatic DNA synthesis technologies are being explored that use template- independent DNA polymerases (TIDPs) for controlled addition of nucleotides to a growing DNA strand. Although advances have been reported recently, enzymatic DNA synthesis is still limited by the low efficiency of available TIDPs, and specifically by the relative inability of these polymerases to incorporate 3'-blocked nucleotides. In this Phase I Small Business Innovation Research (SBIR) project, Primordial Genetics Inc, a synthetic biology company with differentiated combinatorial genetic technology, and Denovium Inc., an artificial intelligence company pioneering novel Al methods for genetic discovery, are joining forces to develop novel and highly efficient TIDPs for enzymatic DNA synthesis in vitro. Denovium will use their computational capabilities based on machine learning algorithms to discover novel TIDPs with the desired activities from proprietary and public databases. Denovium will also perform proprietary artificial intelligence (AI) scans to determine the functional impact of all possible mutations on the selected TIDPs. Primordial Genetics will synthesize and express the resulting collection of sequences, and test them in vitro to identify the most active enzymes. The best 2 enzymes will be diversified using Primordial Genetics' proprietary Function Generator technology and other randomized diversification methods. Populations of genes encoding enzyme variants will be screened with ultra-high-throughput screens to identify the most active enzymes. The dataset resulting from this work will be used to train Denovium's sequence prediction algorithm to accelerate further enzyme improvements in Phase II. The proposed work is a feasibility study for isolating and developing novel enzymes suitable for enzymatic DNA synthesis, and also for creating a pipeline of enzyme optimization tools. The enzymes discovered and in this work will be directly useful for enzymatic DNA synthesis applications, and can be licensed or sold to leading DNA and gene manufaturers.