基本信息
浏览量:214
职业迁徙
个人简介
Research
Protein Function Prediction with Bioinformatics Approaches
The DNA sequence of the human genome is now known, yet we are only starting to understand the functional networks of the proteins encoded in it. The group is addressing this issue with a number of computational approaches, including hidden Markov models, Bayesian networks, clustering algorithms, evolutionary models, and discriminant statistical methods. The goal is to predict function both in terms of biochemical activity and role in a pathway. To predict pathway interactions we are developing integrative computational methods and databases with a focus on identifying new disease genes. The resulting networks are available at http://FunCoup.sbc.su.se/
Research interests
Systems biology: We are developing new methods to make regulatory network inference robust and reliable. The accuracy of regression-based methods depend strongly on a sparsity parameter and we have developed a system to optimise this. We are using the SciLifeLab facilities to perform perturbation experiments by inhibiting gene expression in cell lines and measuring the transcriptomic response. This pipeline is applied to unravel gene regulatory networks implicated in cancer.
Network biology: The FunCoup system builds networks of functionally coupled genes/proteins by integrating data of many types, including microarray mRNA and protein expression, tandem affinity purification, colocalisation, regulatory signals, and evolutionary patterns. Functional coupling observed in one species is transferred using the orthology relations in InParanoid. The data sources are integrated by Bayesian training using up to four gold standard training sets per species, each conveying a different kind of functional link. FunCoup is designed such that multiple evidences for functional coupling add to the total score and give higher confidence scores. The website is the only one that supports ”comparative genomics”, i.e. quering of conserved subnetworks in multiple species (see figure). We are furthermore developing several network analysis tools to identify network modules, measure gene group cross-talk, and to rank new candidate members of a group. The latter was applied to cancer and predicted nearly 2000 novel cancer genes (Östlund et al., 2010).
Orthology identification: we are developing the InParanoid method and database, which is an efficent way to find high-quality orthologs between two species. This algorithm is being upgraded to support multiple species in a hierarchical manner.
Protein domain architecture analysis: We are developing various tools to analyse the evolution of protein domain architectures based on the Pfam database.
研究兴趣
论文共 199 篇作者统计合作学者相似作者
按年份排序按引用量排序主题筛选期刊级别筛选合作者筛选合作机构筛选
时间
引用量
主题
期刊级别
合作者
合作机构
Briefings in bioinformaticsno. 2 (2024)
Frontiers in Genetics (2022)
引用0浏览0引用
0
0
加载更多
作者统计
合作学者
合作机构
D-Core
- 合作者
- 学生
- 导师
数据免责声明
页面数据均来自互联网公开来源、合作出版商和通过AI技术自动分析结果,我们不对页面数据的有效性、准确性、正确性、可靠性、完整性和及时性做出任何承诺和保证。若有疑问,可以通过电子邮件方式联系我们:report@aminer.cn