Uncovering the Human Secretome

项目总结/摘要 肽类激素通过内分泌调节胚胎发育和大多数生理过程 或旁分泌信号。它们也是治疗常见和罕见疾病的相对安全药物的丰富来源 疾病然而，找到低于300个碱基对的肽编码基因本质上是困难的，因为它们位于 基因组的噪音最近的多学科，蛋白系统基因组学研究在较低的物种，如酵母 和苍蝇，已经发现了数百个新的小蛋白质编码基因，称为“smORFs”。在人类中，最近 对线粒体基因组的研究还发现了几十种称为MDP的小肽激素基因。 基于这些和其他研究，估计人类核基因组中约5%的蛋白质具有 尚未发现，特别是那些编码低于100个氨基酸的小肽的。这是一个精心 丢弃大量测序和蛋白质组学数据的做法有文件记录，但很少受到挑战 因为它们与注释的人类基因组不匹配。我的首要目标是发现人类 “分泌物”，并实际利用它来改善人类的状况。在过去的几年里，我们 开发了一个独特的技术管道，结合了数学，计算机硬件和 软件、蛋白质组学、质谱和HTS筛选，每一项都经过优化， 整合我们基于机器学习的GeneBook软件模块可以以100倍的速度处理数据 与传统方法相比，使用公共和内部生成的 基因和蛋白质组数据库。使用平台的原型版本， 在人类、黑猩猩和猕猴之间，我们已经发现了数千个假定的肽编码基因， 验证了数百个。我们的目标是（1）进一步改进算法，以提高其速度和准确性， (2)改进数千个新小基因的基因组注释，（3）确定其表达谱 在正常和患病组织中，（4）探索它们与疾病位点的遗传关联，以及（5）筛选第一个 分泌组文库，以寻找具有新的生物学和治疗相关活性的激素。数据， 将向研究界提供软件包和图书馆。这样做，我们将摆脱 人类基因组的暗物质，具有最大治疗潜力的部分，从而帮助 为子孙后代引导和加快研究和药物开发的步伐。

PROJECT SUMMARY / ABSTRACT Peptide hormones regulate embryonic development and most physiological processes by acting as endocrine or paracrine signals. They are also a rich source of relatively safe medicines to treat both common and rare diseases. Yet finding peptide-coding genes below ~300 base pairs is inherently difficult because they lie within the noise of the genome. Recent multidisciplinary, proteophylogenomic studies in lower species, such as yeast and flies, have uncovered hundreds of new small protein-coding genes called “smORFs”. In humans, recent work on the mitochondrial genome has also uncovered dozens of small peptide hormone genes called MDPs. Based on these and other studies, it is estimated that about 5% of proteins in the human nuclear genome have not yet been discovered, particularly those that encode small peptides below 100 amino acids. It is a well documented but rarely challenged practice to discard large quantities of sequencing and proteomic data because they do not match the annotated human genome. My overarching goal is to discover the human “secretome” and make practical use of it to improve the human condition. Over the past few years, we have developed a unique pipeline of technologies that combines breakthroughs in math, computer hardware and software, proteomics, mass spectrometry, and HTS screening, each of which has been optimized and integrated. Our GeneFinder software modules, based on machine-learning, can process data 100 times faster than traditional methods and rapidly validate small human genes using public and in-house generated databases of genetic and proteomic data. Using the prototype version of the platform that finds conservation between humans, chimp, and macaque, we have discovered thousands of putative peptide-coding genes and validated hundreds of them. We aim to (1) further improve the algorithm to increase its speed and accuracy, (2) improve the genome annotation for thousands of small novel genes, (3) determine their expression profiles in normal and diseased tissues, (4) explore their genetic association with disease loci, and (5) screen the first secretomic library to find hormones with novel biological and therapeutically relevant activities. The data, the software package, and libraries will be made available to the research community. In doing so, we will shed light on the dark matter of the human genome, the parts with the greatest therapeutic potential, thereby helping to steer and accelerate the pace of research and drug development for generations to come.