Transcription Factor-DNA Interactions: Structural Classification and Prediction

描述（由申请人提供）： 人类疾病通常是由生物体中特定基因的过度或缺陷转录引起的。被称为转录因子（TF）的蛋白质通过结合DNA上被称为TF结合位点（TFBS）的特异性位点来调节基因的转录。用于研究TF如何识别TFBS的策略之一是通过开发用于预测基因组序列中TFBS的计算机算法。大多数算法应用从TFBS的DNA序列获得的位置特定的权重矩阵（PWM）。然而，对于在人类基因组中已经识别和预测的3,000多个TF，仅为300个TF构建了PWM。为了将预测方法应用于具有未知PWM的TF，应用两种方法。第一个建议产生PWM的家庭的TF共享类似的DNA结合域。然而，来自相同家族的TF，如在TF的现有分类中指定的，通常不识别相同的DNA序列。因此，存在对能够驱动TFBS的预测的TF的新分类的呼吁。第二种方法是在TF-DNA复合物的三维（3D）结构上构建PWM。正如我们的初步结果表明，这种方法也可以适用于家庭的TF的PWM可以通过TFBS序列和TF-DNA复合物的3D结构的TF家族的比对。这样产生的PWMs TF家庭的性能建议用于验证TF的分类。该提案的目标是研究TF的DNA结合结构域的序列和结构的相似性如何与TFBS的相似性相关。研究将集中在TF的至少一个三维结构的TF-DNA复合物或其密切的同系物是可用的。本研究的主要目标是：（1）建立一种基于TFs和TFBS序列和结构相似性的TFs DNA结合域的自动分类方法;（2）建立一种基于结构的TFBS预测方法;以及（3）通过网络资源传播结果，该网络资源以查询和网络工具的形式提供对分类和预测方法的访问。研究结果将对人类和其他生物基因组中转录因子和调控区的注释具有重要意义。它们还将有助于破译基因调控网络，并设计用于治疗与基因调控不足相关的疾病的药物。这项研究的结果将有助于注释人类和其他生物基因组中的转录因子和调控区域。这些信息对于设计用于治疗与基因调控不足相关的疾病的药物具有重要意义。

DESCRIPTION (provided by applicant): Human diseases often arise from excessive or deficient transcription of particular genes in an organism. Proteins known as transcription factors (TF) regulate transcription of a gene through binding specific sites on DNA known as TF binding sites (TFBS). One of the strategies used to study how TFs recognize TFBSs is through the development of computer algorithms for predicting TFBSs in genomic sequence. Most algorithms apply position-specific weight matrices (PWM) obtained from DNA sequences of TFBSs. However, for more than 3,000 TFs that have been identified and predicted in the human genome, PWMs were built for only 300 TFs. To apply the prediction methods to TFs with unknown PWMs, two approaches are applied. The first one proposes generating PWMs for families of TFs sharing similar DNA-binding domains. However, TFs from the same family, as designated in the existing classifications of TFs, often do not recognize the same DNA sequences. Hence, there is an appeal to a new classification of TFs able to drive the prediction of TFBSs. The second approach is to build PWMs on three- dimensional (3D) structures of TF-DNA complexes. As our preliminary results indicate, this approach can also be applied to families of TFs a PWM can be obtained for a TF family through alignment of TFBS sequences and 3D structures of TF-DNA complexes. The performance of such generated PWMs for TF families is suggested to be used for validation of the classification of TFs. The goal of the proposal is to study how similarity in sequences and structures of DNA-binding domains of TFs relates to the similarity of TFBSs. The research will focus on TFs for which at least one 3D structure of TF-DNA complex or its close homolog is available. The goal will be attained through the following specific aims: (1) Develop an automatic classification of DNA-binding domains of TFs based on similarity of sequences and structures of TFs and TFBSs; (2) Develop a structure-based approach to the prediction of TFBSs for families of TFs; and (3) Disseminate the results by means of a web resource providing access to the classification and prediction methods in the form of queries and web tools. The results of the study will be valuable for annotating TFs and regulatory regions in genomes of human and other organisms. They will also facilitate deciphering of gene regulatory networks and designing drugs for treatment of diseases associated with inadequate gene regulation. The results of the proposed study will facilitate annotating transcription factors and regulatory regions in genomes of human and other organisms. This information is significant for designing drugs for treatment of diseases associated with inadequate gene regulation.