转录因子作为一种重要的转录调控因子，通过与特异的DNA序列结合，调控下游基因的表达。随着测序技术的发展，ChIP-seq已经成为研究转录因子和DNA结合关系的重要实验手段。如何分析ChIP-seq数据，确定转录因子的特异识别序列，寻找转录因子的特异调控基因，对于理解转录因子的功能具有重要科学意义。现在用于寻找ChIP-seq中的转录因子特异识别序列的模体识别算法存在计算速度缓慢、依赖重复序列过滤、无法区分识别位点真伪等缺点。本项目拟在已发表的ChIP-seq数据基础上，挖掘和总结ChIP-seq数据的特点，针对这些特点，利用朴素贝叶斯分类器、最大似然法和boosting算法，开发新的模体识别算法，从而改善目前算法存在的缺点，大大提高算法对ChIP-seq数据分析的针对性和整体性能。这些研究结果，将全面提高ChIP-seq数据模体识别算法的性能，对转录因子的研究提供新的技术手段和重要工具。

Transcription factors are crucial components for gene regulation. They control the downstream gene expression by interacting with specific DNA elements. With the development of sequencing technologies, ChIP-seq has become one of the most important strategies to investigate the relationship between transcription factors and DNA binding sites. Therefore, how to analyze ChIP-seq data and to identify the specific DNA elements have become one important scientific question. Several motif search algorithms have been developed and applied to identify the specific DNA elements in ChIP-seq data, however, current methods have a few drawbacks, including slow, relying on repeat masking and difficulties in distinguishing real binding sites. Here we propose to develop new motif search algorithms based on na?ve Bayes classifier, maximum likelihood and boosting algorithm and by digging and summarizing the special characteristics of ChIP-seq data. The new algorithm will overcome the drawbacks of old algorithms and will be more efficient for identifying specific targets of transcription factors. Our results will greatly improve the performance of motif search algorihtms on ChIP-seq data and will provide an important tool for studying transcription factors.