SBIR Phase I: Genomics Accelerator

这个小型企业创新研究（SBIR）项目的更广泛的影响/商业潜力是通过开发精准医疗的计算解决方案来彻底改变医疗保健行业。 下一代测序（NGS）现在以可承受的成本产生大量的基因组数据。分析来自不同人群的大量基因组数据将有助于更好地了解从癌症到罕见遗传疾病等各种疾病的原因，并开发出更好的治疗方法。通过对患者的基因组进行测序，可以更具体地确定他/她的病情并制定个性化的治疗计划。而且，如果每个新生儿的基因组都被测序，它将允许主动识别疾病风险和早期干预。了解病毒和细菌将有助于我们更好地了解传染病，从而使我们能够开发更好的免疫和疾病控制方法。除了医疗保健，基因组学还可以帮助设计更好的粮食作物和实现粮食安全。 该项目将开发为基因组测序软件定制的新型计算硬件解决方案。这些计算解决方案可以帮助实现数字基因组学市场的指数增长，并为所有人带来负担得起的个性化医疗。SBIR第一阶段项目提出通过构建基因组自动机的定制处理器来加速全基因组测序（WGS）。 在过去的十年里，测序的生产成本从一千万美元下降到一千美元，预计很快就会低于每个基因组的一百美元。然而，处理器性能的进步并没有跟上步伐。一个基因组的排序可能需要几天的时间。为了使基因组测序变得像常规医学测试一样负担得起，需要能够在几分钟内对基因组进行测序的计算解决方案，并且每个基因组的成本仅为几美元。该项目正在开发一种用于计算有限状态自动机的新型ASIC硬件加速器，该加速器支持近似字符串匹配，这是一种广泛用于测序等基因组学应用的计算内核。该设计是基于一种新的内存中的硬件技术，有效地实现状态转换。目标是将基因组自动机的效率与现代处理器相比提高两个数量级。

The broader impact/commercial potential of this Small Business Innovation Research (SBIR) project is to revolutionize the healthcare industry by developing computing solutions for precision medicine. Next-Generation Sequencing (NGS) now produces a large amount of genome data at an affordable cost. Analyzing large volumes of genome data from diverse populations will lead to a better understanding of the causes of various diseases ranging from cancer to rare genetic disorders, and develop better cures. By sequencing a patient's genome, it is possible to more specifically determine his/her condition and devise an individualized treatment plan. And, if every newborn's genome was sequenced, it would allow for proactive identification of disease risks, and early intervention. Understanding viruses and bacteria will help us gain a better understanding of infectious diseases, and thereby enable us to develop better immunization and disease control methods. In addition to health-care, genomics also may help engineer better food crops and attain food security. This project will develop novel computing hardware solutions that are customized for genomics sequencing software. These computing solutions can help realize an exponential growth in the digital genomics market and bring affordable personalized medicine to all.This SBIR Phase I project proposes to accelerate Whole Genome Sequencing (WGS) by building a customized processor for genomics automata. Over the last decade, the production cost of sequencing has plummeted from ten million dollars to a thousand dollars, and is soon expected to go below a hundred dollars per genome. Advancements in processor performance, however, have not kept pace. It can take several days on a computing cluster to sequence a genome. For genome sequencing to become as affordable as a routine medical test, computing solutions are required that can sequence a genome in minutes, and cost only a few dollars per genome. This project is developing a novel ASIC hardware accelerator for computing finite state automata, which supports approximate string matching, a computational kernel widely used in genomics applications such as sequencing. The design is based on a novel in-memory hardware technology for efficiently implementing state transitions. The goal is the improve efficiency of genomics automata by two orders of magnitude compared to modern processors.