SBIR PHASE I - TOPIC 407 - CLOUD-BASED SOFTWARE FOR THE CANCER RESEARCH DATA COMMONSPERIOD OF PERFORMANCE: 09/16/2020-06/15/2021

鸟枪式蛋白质组学是理解复杂人类样本蛋白质组最常用的技术。在NCI的蛋白质组学云上有数TB的蛋白质组学数据可用，但不幸的是，大多数蛋白质组学搜索引擎搜索规范的人类序列数据库以及 一把PTM。因此，50%以上的MS/MS光谱仍未得到解释。我们开发了一种基于云的算法Bolt，它可以在几分钟内搜索近240万个蛋白质序列(规范、异构体和突变)以及41个PTM和部分胰酶搜索。Bolt能够在几分钟内搜索到这个50倍大的搜索空间，而传统搜索引擎即使在较小的搜索空间上也需要更长的时间。利用云计算的力量，我们能够比传统搜索引擎多排序20%的MS/MS。我们建议将Bolt调整和集成到NCI的云平台，并将其连接到CRDC蛋白质组学管道(用于鸟枪式蛋白质组学和新抗原发现)，在那里科学家可以无缝地分析他们现有和未来的数据集，同时比以前更深入地挖掘它们。我们还建议连接基因组节点以提取样本特定信息，以进一步挖掘新编码区的蛋白质组数据集。

Shotgun proteomics is the most commonly used technique for understanding the proteome of complex human samples. There is terabytes of proteomics data available on NCI's proteomics cloud, but unfortunately most proteomics search engine search a canonical human sequence database along with a handful of PTMs. Thus more than 50% of the MS/MS spectra remain uninterpreted. We have developed a cloud-based algorithm, Bolt, that can search almost more than 2.4 million protein sequences (canonical, isoforms and mutations) along with 41 PTMs and partial tryptic search in a matter of minutes. Bolt is able to search this 50 times larger search space in minutes whereas the traditional search engine need much longer even on a smaller search space. Utilizing the power of cloud computing, we are able to sequence 20% more MS/MS than traditional search engines. We propose adapting and integrating Bolt to the NCI's cloud platform, and linking it to the CRDC proteomics pipeline (for shotgun proteomics and neoantigen discovery) where scientists can analyse their existing and future data sets seamlessly while mining them much deeper than before. We also propose to link the genomic node to extract sample-specific information to futher mine proteogenomics data set for novel coding regions.