The systematic functional analysis of regulatory novel open reading frames.

近93%的疾病和性状相关的单核苷酸多态(SNPs)被映射到非编码区。为了更好地了解非编码区的致病性，我们引入了新的开放阅读框架(NORFs)基因类，并研究了非规范转录和翻译产物的功能同一性。目前，Norf数据库OpenProt提供了143,102个蛋白质的基于质谱学的翻译证据。尽管有确凿的证据证明Norf的翻译和调控，但对Norf的功能和疾病关联的全面理解仍然具有挑战性。然而，生物信息学和数据获取方面的最新进展有望为开发数据驱动的管道提供新的机会，从而可能揭示尚未被忽视的(非正则)蛋白质的身份和调节功能。在这份报告中，我们分享了初步的结果，并提出了一项计划，以进一步建立nORF与生物信息学分析和实验验证的相关性，以更好地了解广泛的疾病。本报告首先对各领域、相关性和进展情况作了一般性介绍，随后又分成4章，每一章都有助于对北大西洋公约组织的候选人进行严格的自上而下的分析和确定优先次序。在第一章中，我们介绍了Norf数据库和平台(nORFs.org)的创建，从那时起，它就构成了所有后续研究的参考数据库。第二章利用所创建的数据库来设计特征描述符尺寸，以供以后的分析管道使用。第三章介绍了一种用于质量控制和分析nORF、RefSeq基因和随机序列的组合数据集的高维分析管道。在分析中，我们使用非监督聚类方法(t-SNE和UMAP)作为复杂机器学习方法的基础。在第四章中，从t-SNE簇中提取了37个序列样本，以确定潜在的功能关系。在这些分析步骤之后，报告总结了对Norf监管的初步见解和未来的研究考虑。(万岁节选)

Nearly 93% of all disease and trait associated single nucleotide polymorphisms (SNPs) are mapped to non-coding regions. To gain a better understanding of the pathogenicity of non-coding regions, we here introduce the gene class of novel open reading frames (nORFs) and investigate the functional identity of non-canonical transcriptional and translational products. At the time being, the nORF database openProt provides mass spectroscopy based translational evidence of 143,102 proteins. Despite decisive evidence of nORF translation and regulation, the comprehensive understanding of nORF function and disease-association remains to be challenging. However, recent advances in bioinformatics and data acquisition promise new opportunities to develop data-driven pipelines and might thus unveil the identity and regulatory function of yet overlooked (non-canonical) proteins. In this report, we share preliminary results and propose a plan to further establish the relevance of nORFs with the means of bioinformatic analysis and experimental validation to gain a better understanding of a wide array of diseases. This report begins with a general introduction of the fields relevance and progress followed up by 4 chapters that each contribute to the rigorous top-down analysis and prioritization of functional nORF candidates. In chapter one we introduce the creation of the nORF database and platform (nORFs.org) that from then on constitutes as reference database for all later research. Chapter two utilizes the created database to engineer feature descriptor dimensions for later analysis pipelines. Chapter three introduces a high dimensional analysis pipeline to quality control and analyze a combined dataset of nORFs, RefSeq genes and random sequences. In the analysis, we utilize unsupervised clustering methods (t-SNE and UMAP) as groundwork for sophisticated machine learning methods. In chapter four 37 sequence samples were taken from t-SNE clusters to identify potential functional relationships. Subsequent to those analysis steps the report concludes with preliminary regulatory nORF insights and future research considerations. (VIVA excerpt)