SHF: Medium: Custom Computing for Genome Sequencing

基因组学有可能通过简单的血液测试实现癌症的早期检测，治疗罕见的遗传疾病，识别传染性病原体等，从而彻底改变精准健康。在临床实践中采用基因组测序严重依赖于我们分析大量数据的计算能力，这些数据比现代数据中心处理的数据要大得多。但不幸的是，由于摩尔定律即将结束，通用处理器性能的增长停滞不前，因此商品系统不太可能满足基因组学中不断增长的计算需求。为了满足这一需求，该项目正在开发与生物信息学测序软件共同设计和集成的定制硬件加速器，以实现基因组测序的广泛使用。该项目的软件和硬件工件和基准可以直接影响基因组学从业者，以及刺激该领域的系统研究。该项目还支持广泛的教育和外展活动。该项目有三个具体目标。设计面向全基因组测序（WGS）的领域动态规划和内存加速器。它们将支持广泛的基因组学内核，包括新兴的长读测序和泛基因组学，而不是特定于内核的加速器。第二，使用牛津纳米孔技术公司（ONT）的长读实时测序仪开发即时病原体检测器。该项目将开发定制的硬件解决方案，以实时分析测序的原始数据（曲线），利用ONT测序仪的读取至功能，并在护理点产生可操作的诊断结果。第三，术中液体和组织活检的实时测序，可以在几分钟到几小时内得出结果，而不是现在需要几天到几周的时间。该奖项反映了美国国家科学基金会的法定使命，并通过使用基金会的知识价值和更广泛的影响审查标准进行评估，被认为值得支持。

Genomics has the potential to revolutionize precision health by enabling the early detection of cancer through simple blood tests, treating rare genetic diseases, identifying infectious pathogens, and more. Adoption of genome sequencing in clinical practice is critically dependent on our computational ability to analyze large volumes of data – significantly larger than data processed today in modern data-centers. But unfortunately, due to the impending end of Moore’s law, growth in general-purpose processor performance has stagnated, and therefore commodity systems are unlikely to meet the growing computational needs in genomics. To address this need, this project is developing custom hardware accelerators that are co-designed and integrated with bioinformatics sequencing software to enable pervasive use of genome sequencing. Software and hardware artifacts and benchmarks from this project can directly impact genomics practitioners, as well as spur systems research in this area. This project also supports a broad range of educational and outreach activities.The project has three specific aims. One, design domain-specific dynamic programming and in-memory accelerators for whole genome sequencing (WGS). Instead of kernel-specific accelerators, they will support a wide range of genomics kernels, including emerging long-read sequencing and pangenomics. Two, develop point-of-care pathogen detectors using Oxford Nanopore Technologies' (ONT) long-read real-time sequencers. The project will develop custom hardware solutions to analyze sequenced raw data (squiggles) in real-time to leverage ONT sequencer’s read-until feature and produce actionable diagnostics results at point-of-care. Three, real-time sequencing for intra-operative liquid and tissue biopsy that can produce results in minutes to hours, as opposed to today where it takes days to weeks.This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.