Hackathon - TF-TG literature triage Gold Standard corpus

This link contains the gold standard corpus for training the systems aiming at classifying or triaging the documents (PubMed abstracts) in relevant or not relevant for TF-TG relations. Note that this is not purely a curation triage, but we are primarily looking whether the abstracts do describe/characterize TF-TG interactions (positive training set labeled as “1” or they don't (negative training set labeled as “0”). This dataset was manually classified by domain experts into these to classes. Note that in PubMed there is a very considerable class imbalance, implying that there is only a very small fraction of TF-TG relations relevant literature subset. The resulting classifier should be useful to triage the entire PubMed database. Name: greekc_triage_training_v02.tsv Example: 11731616 [{"sourceid":"11731616","sourcedb":"PubMed","text":"Homeobox protein Gsh-1-dependent regulation of the rat GHRH gene promoter. Although GHRH is known to play a pivotal role in the regulation of the GHRH-GH-IGF-I axis, the molecular mechanism of GHRH gene expression has not yet been examined. Here we studied the transcriptional regulation of the GHRH gene 5'promoter using an in vitro experimental model system. We especially focused on the role of homeobox transcriptional factor Gsh-1, because a dwarf phenotype and abolished GHRH expression was observed in Gsh-1 knockout mice. First, we cloned human Gsh-1, which showed 87.3% homology with mouse Gsh-1 at the nucleotide level. When the 5'-promoter region of the rat GHRH gene was introduced into the human placental cell line JEG-3, in which we found the endogenous expression of Gsh-1 as well as GHRH mRNA, substantial transcriptional activity of the promoter was recognized. Promoter activity was further enhanced by overexpression of Gsh-1 protein, whereas it was substantially reduced by elimination of Gsh-1 binding sites. EMSA confirmed the actual binding of Gsh-1 on the multiple binding sites of GHRH gene promoter. Finally, coexpression of CREB-binding protein significantly enhanced the Gsh-1-induced GHRH gene expression, suggesting the cooperative role of the coactivator protein. Because Gsh-1 is found to be expressed in the hypothalamus of the adult rat, our data provide evidence that the Gsh-1 homeobox protein plays a key role in the expression of the GHRH gene.","denotations":[{"obj":"Gene:29446","span":{"begin":84,"end":88}},{"obj":"Gene:288457","span":{"begin":17,"end":22}},{"obj":"Gene:29446","span":{"begin":55,"end":59}},{"obj":"Gene:29446","span":{"begin":146,"end":150}},{"obj":"Gene:29446","span":{"begin":193,"end":197}},{"obj":"Gene:29446","span":{"begin":295,"end":299}},{"obj":"Gene:14842","span":{"begin":430,"end":435}},{"obj":"Gene:14601","span":{"begin":477,"end":481}},{"obj":"Gene:14842","span":{"begin":509,"end":514}},{"obj":"Gene:219409","span":{"begin":553,"end":558}},{"obj":"Gene:14842","span":{"begin":599,"end":604}},{"obj":"Gene:29446","span":{"begin":669,"end":673}},{"obj":"Gene:219409","span":{"begin":783,"end":788}},{"obj":"Gene:2691","span":{"begin":800,"end":804}},{"obj":"Gene:288457","span":{"begin":940,"end":945}},{"obj":"Gene:288457","span":{"begin":1010,"end":1015}},{"obj":"Gene:288457","span":{"begin":1068,"end":1073}},{"obj":"Gene:29446","span":{"begin":1107,"end":1111}},{"obj":"Gene:54244","span":{"begin":1152,"end":1172}},{"obj":"Gene:288457","span":{"begin":1200,"end":1205}},{"obj":"Gene:29446","span":{"begin":1214,"end":1218}},{"obj":"Gene:288457","span":{"begin":1304,"end":1309}},{"obj":"Gene:288457","span":{"begin":1408,"end":1413}},{"obj":"Gene:29446","span":{"begin":1473,"end":1477}}]}] 1 Format: tsv-separated columns (PMID, PubAnnotation JSON formated results of Pubtator for this record together with the automatically detected gene mentions using GnormPlus providing the Entrez Gene Identifiers together with the mention offsets, i.e. start and end character positions, corresponding class label of the records (1= relevant, 0=nonrelevant)) PubAnnotation format description: http://www.pubannotation.org/docs/annotation-format/ PubTator record retrieval description: https://www.ncbi.nlm.nih.gov/CBBresearch/Lu/Demo/tmTools/curl.html <strong>Remember:</strong> One must notice that this dataset is high imbalanced, with very few cases of the positive class. The additional files could also be explored to build the model or even to build a classifier working at the level of sentences instead of abstracts,本链接提供用于训练相关系统的金标准语料库，该系统旨在对文献（PubMed摘要）进行分类或分诊，以判定其是否与TF-TG（转录因子-靶基因，Transcription Factor-Target Gene）关系相关。请注意，本任务并非单纯的人工整理分诊，我们的核心目标是判定摘要是否描述或阐明了TF-TG相互作用：符合该条件的正训练集标注为“1”，不符合的负训练集标注为“0”。 本数据集由领域专家手动标注为上述两类。 需注意，PubMed数据库中存在显著的类别不平衡问题，即与TF-TG关系相关的文献子集占比极低。本数据集训练得到的分类器可用于对整个PubMed数据库进行分诊筛选。 数据集名称：greekc_triage_training_v02.tsv 示例： 11731616 [{"sourceid":"11731616","sourcedb":"PubMed","text":"Homeobox蛋白Gsh-1对大鼠GHRH基因启动子的依赖性调控。尽管已知GHRH在GHRH-GH-IGF-I轴的调控中发挥关键作用，但GHRH基因表达的分子机制尚未得到研究。本研究利用体外实验模型系统探究了GHRH基因5'启动子的转录调控机制。我们特别聚焦于同源盒转录因子Gsh-1的作用，因为在Gsh-1敲除小鼠中观察到了侏儒表型和GHRH表达缺失。首先，我们克隆了人源Gsh-1，其核苷酸序列与小鼠Gsh-1的同源性达87.3%。将大鼠GHRH基因的5'-启动子区域导入人胎盘细胞系JEG-3（我们在该细胞系中检测到Gsh-1和GHRH mRNA的内源性表达）后，观察到该启动子具有显著的转录活性。过表达Gsh-1蛋白可进一步增强启动子活性，而去除Gsh-1结合位点则会显著降低其活性。电泳迁移率变动分析（EMSA，Electrophoretic Mobility Shift Assay）证实了Gsh-1确实结合在GHRH基因启动子的多个结合位点上。最后，共表达CREB结合蛋白可显著增强Gsh-1诱导的GHRH基因表达，提示该辅激活蛋白具有协同作用。由于已知成年大鼠下丘脑组织中表达Gsh-1，本研究数据证明同源盒蛋白Gsh-1在GHRH基因的表达中发挥关键作用。","denotations":[{"obj":"Gene:29446","span":{"begin":84,"end":88}},{"obj":"Gene:288457","span":{"begin":17,"end":22}},{"obj":"Gene:29446","span":{"begin":55,"end":59}},{"obj":"Gene:29446","span":{"begin":146,"end":150}},{"obj":"Gene:29446","span":{"begin":193,"end":197}},{"obj":"Gene:29446","span":{"begin":295,"end":299}},{"obj":"Gene:14842","span":{"begin":430,"end":435}},{"obj":"Gene:14601","span":{"begin":477,"end":481}},{"obj":"Gene:14842","span":{"begin":509,"end":514}},{"obj":"Gene:219409","span":{"begin":553,"end":558}},{"obj":"Gene:14842","span":{"begin":599,"end":604}},{"obj":"Gene:29446","span":{"begin":669,"end":673}},{"obj":"Gene:219409","span":{"begin":783,"end":788}},{"obj":"Gene:2691","span":{"begin":800,"end":804}},{"obj":"Gene:288457","span":{"begin":940,"end":945}},{"obj":"Gene:288457","span":{"begin":1010,"end":1015}},{"obj":"Gene:288457","span":{"begin":1068,"end":1073}},{"obj":"Gene:29446","span":{"begin":1107,"end":1111}},{"obj":"Gene:54244","span":{"begin":1152,"end":1172}},{"obj":"Gene:288457","span":{"begin":1200,"end":1205}},{"obj":"Gene:29446","span":{"begin":1214,"end":1218}},{"obj":"Gene:288457","span":{"begin":1304,"end":1309}},{"obj":"Gene:288457","span":{"begin":1408,"end":1413}},{"obj":"Gene:29446","span":{"begin":1473,"end":1477}}]}] 1 数据格式：采用制表符分隔的列，各列依次为：PubMed ID（PMID）、本记录的PubTator结果（采用PubAnnotation JSON格式）、通过GnormPlus自动识别的基因提及内容（包含Entrez Gene标识符及提及的字符偏移量，即起始和结束字符位置），以及该记录对应的类别标签（1=相关，0=不相关） PubAnnotation格式说明：http://www.pubannotation.org/docs/annotation-format/ PubTator记录检索说明：https://www.ncbi.nlm.nih.gov/CBBresearch/Lu/Demo/tmTools/curl.html 注意事项：本数据集存在严重的类别不平衡问题，正例样本占比极低。此外，还可利用额外文件构建模型，甚至构建以句子而非摘要为处理单元的分类器。