Assessment of disorder predictions in CASP7
Assessment of disorder predictions in CASP7
复制标题
DOI：
10.1002/prot.21671
复制
发表时间：
2007-01-01
影响因子：
2.9
通讯作者：
Schwede, Torsten
中科院分区：
文献类型：
--
作者：
Bordoli, Lorenza;Kiefer, Florian;Schwede, Torsten
蛋白质中的非结构化区域与许多重要的生物细胞功能相关。由于通过实验测量天然无序在技术上具有挑战性，因此预测蛋白质无序区域的计算方法近年来引起了人们的极大兴趣。作为第七次蛋白质结构预测技术关键评估（CASP7）的一部分，我们评估了19种基于96种靶蛋白结果的疾病预测方法。预测精度进行了评估，使用详细的数值比较预测的混乱和实验结构。平均而言，参与CASP7的方法与CASP6的先前评估相比提高了准确性。然而，总体而言，在CASP7中未观察到CASP6中最佳方法的改善。不同预测方法在基于蛋白质靶序列正确预测有序和无序残基的灵敏度和特异性方面存在显着差异，这与这些计算工具的实际应用相关。
Intrinsically unstructured regions in proteins have been associated with numerous important biological cellular functions. As measuring native disorder experimentally is technically challenging, computational methods for prediction of disordered regions in a protein have gained much interest in recent years. As part of the seventh Critical Assessment of Techniques for Protein Structure Prediction (CASP7), we have assessed 19 methods for disorder prediction based on their results for 96 target proteins. Prediction accuracy was assessed using detailed numerical comparison between the predicted disorder and the experimental structures. On average, methods participating in CASP7 have improved accuracy in comparison to the previous assessment in CASP6. Overall, however, no improvement over the best methods in CASP6 was observed in CASP7. Significant differences between different prediction methods were identified with regard to their sensitivity and specificity in correctly predicting ordered and disordered residues based on a protein target sequence, which is of relevance for practical applications of these computational tools.