NMDetective

# Genome-wide predictions of NMD efficacy described in Lindeboom et al, Nature Genetics, 2019.<br>The NMDetective resource predicts the efficacy of nonsense-mediated mRNA decay for a pre-mature stop codon at every protein-coding nucleotide in the human and mouse genome. NMD scores range from ~0, indicating no mRNA degradation by NMD, to &gt;0.52, indicating strong mRNA degradation. NMDetectiveA is a comprehensive prediction of NMD efficacy by Random Forest regression. NMDetectiveB is a simplified and fully transparent decision tree model that predicts NMD efficacy based on the EJC-model, distance to the start codon, and the length of the affected exon.<br>This fileset contains the NMDetectiveA and B predictions on the UCSC knownGene databases of the hg38, hg19, mm10 and mm9 genome builds in GTF/GFF2.2 format. The NMD scores are added as an attribute to each protein-coding exon as comma separated values per nucleotide.<br>Non-overlapping bedGraphs of the NMDetectiveA and NMDetectiveB predictions are also accessible and visualized in the following UCSC genome browser sessions:<br>http://genome-euro.ucsc.edu/cgi-bin/hgTracks?hgS_doOtherUser=submit&amp;hgS_otherUserName=Rik Lindeboom&amp;hgS_otherUserSessionName=hg19_NMDetective <br>http://genome-euro.ucsc.edu/cgi-bin/hgTracks?hgS_doOtherUser=submit&amp;hgS_otherUserName=Rik Lindeboom&amp;hgS_otherUserSessionName=hg38_NMDetective <br>http://genome-euro.ucsc.edu/cgi-bin/hgTracks?hgS_doOtherUser=submit&amp;hgS_otherUserName=Rik Lindeboom&amp;hgS_otherUserSessionName=mm9_NMDetective <br>http://genome-euro.ucsc.edu/cgi-bin/hgTracks?hgS_doOtherUser=submit&amp;hgS_otherUserName=Rik Lindeboom&amp;hgS_otherUserSessionName=mm10_NMDetective<br><br>

# 本数据集为Lindeboom等人2019年发表于《Nature Genetics》的全基因组无义介导mRNA降解（Nonsense-mediated mRNA decay, NMD）效率预测结果。
NMDetective资源可针对人类和小鼠基因组中每个蛋白质编码核苷酸处的提前终止密码子，预测其对应的NMD效率。NMD得分范围约为0（表示无NMD介导的mRNA降解）至>0.52（表示高效mRNA降解）。其中NMDetectiveA是基于随机森林（Random Forest）回归的NMD效率综合预测模型；NMDetectiveB是简化且完全透明的决策树（Decision Tree）模型，其基于外显子连接复合物（Exon Junction Complex, EJC）模型、距起始密码子的距离以及受影响外显子的长度来预测NMD效率。
本文件集包含针对hg38、hg19、mm10及mm9基因组组装版本的UCSC knownGene数据库，以GTF/GFF2.2格式输出的NMDetectiveA和B预测结果。NMD得分以单核苷酸为单位的逗号分隔数值形式，作为属性添加至每个蛋白质编码外显子。
NMDetectiveA与NMDetectiveB预测结果的非重叠bedGraph文件亦可获取，并可通过以下UCSC基因组浏览器会话进行可视化展示：
http://genome-euro.ucsc.edu/cgi-bin/hgTracks?hgS_doOtherUser=submit&hgS_otherUserName=Rik Lindeboom&hgS_otherUserSessionName=hg19_NMDetective
http://genome-euro.ucsc.edu/cgi-bin/hgTracks?hgS_doOtherUser=submit&hgS_otherUserName=Rik Lindeboom&hgS_otherUserSessionName=hg38_NMDetective
http://genome-euro.ucsc.edu/cgi-bin/hgTracks?hgS_doOtherUser=submit&hgS_otherUserName=Rik Lindeboom&hgS_otherUserSessionName=mm9_NMDetective
http://genome-euro.ucsc.edu/cgi-bin/hgTracks?hgS_doOtherUser=submit&hgS_otherUserName=Rik Lindeboom&hgS_otherUserSessionName=mm10_NMDetective

搜集汇总

背景与挑战

背景概述

NMDetective是一个基因组范围的数据集，用于预测人类和小鼠基因组中每个蛋白质编码核苷酸的无义介导mRNA衰变（NMD）功效，分数范围从0（无降解）到大于0.52（强降解）。它包含两个模型：NMDetectiveA（基于随机森林回归的综合预测）和NMDetectiveB（基于透明决策树的简化模型），数据以GTF/GFF2.2格式提供，覆盖多个基因组构建，并支持UCSC基因组浏览器可视化，适用于癌症免疫治疗、CRISPR设计和遗传疾病等研究领域。