期刊
IEEE-ACM TRANSACTIONS ON COMPUTATIONAL BIOLOGY AND BIOINFORMATICS
卷 12, 期 5, 页码 1137-1145出版社
IEEE COMPUTER SOC
DOI: 10.1109/TCBB.2015.2396057
关键词
DNA melting bubbles; structure prediction; correlation and regression analysis; machine learning
类别
资金
- European Union [282881, 261810]
- European Social Fund of the European Union [P-MR-09/124]
- Slovenian Research Agency [P1-0237]
DNA melting bubbles are the basis of many DNA-protein interactions, such as those in regulatory DNA regions driving gene expression, DNA replication and bacterial horizontal gene transfer. Bubble formation is affected by DNA duplex stability and thermally induced duplex destabilization (TIDD). Although prediction of duplex stability with the nearest neighbor (NN) method is much faster than prediction of TIDD with the Peyrard-Bishop-Dauxois (PBD) model, PBD predicted TIDD defines regulatory DNA regions with higher accuracy and detail. Here, we considered that PBD predicted TIDD is inherently related to the intrinsic duplex stabilities of destabilization regions. We show by regression modeling that NN duplex stabilities can be used to predict TIDD almost as accurately as is predicted with PBD. Predicted TIDD is in fact ascribed to non-linear transformation of NN duplex stabilities in destabilization regions as well as effects of neighboring regions relative to destabilization size. Since the prediction time of our models is over six orders of magnitude shorter than that of PBD, the models present an accessible tool for researchers. TIDD can be predicted on our webserver at http://tidd.immt.eu
作者
我是这篇论文的作者
点击您的名字以认领此论文并将其添加到您的个人资料中。
推荐
暂无数据