Scholars at Nankai University have Achieved Excellent Marks in the Critical Assessment of Protein Intrinsic Disorder Prediction (CAID)

2021-05-07


In recent days, Nature Methods has published a research paper entitled Critical Assessment of Protein Intrinsic Disorder Prediction and posted news entitled A Community Effort to Bring Structure to Disorder, which reported the details of the Critical Assessment of Protein Intrinsic Disorder Prediction (CAID). Hu Gang, a professor at the School of Statistics and Data Science, Wang Kui, an associate professor at the School of Mathematical Sciences, and Lukasz Kurgan, a professor at Virginia Commonwealth University in the United States, jointly developed the prediction method fIDPnn, which performed well in this competition and ranked the highest in many indicators.


Figure 1:process and results of CAID


It is generally believed that proteins fold into defined structures to perform functions. However, recent studies have shown that there are some disordered regions in proteins that do not have defined structures. These regions play a key role in the interaction between proteins and other molecules, participating in important cell functions such as signaling, transcription, and translation. However, due to the difficulty of identifying protein structure through experimental methods, computational prediction methods remain essential for studying unstructured regions. This competition is the first international competition for protein intrinsic disorder prediction, attracting dozens of international teams engaged in research on protein disorder regions.


Figure 2:Comparison of Performance metrics between fIDPnn and other methods


This competition has undertaken a detailed evaluation of dozens of methods submitted by participants. FIDPnn, developed by Professor Hu Gang et al., is based on machine learning to accurately predict disordered regions in combination with various characteristics of protein sequences. It has won first place under Disprot standard and Disordered Sequence Prediction, and is also among the best under Disprot-PDB Standard. FIDPnn is not only a leader in prediction accuracy, but also runs more than 10 times faster than its competitors, which greatly facilitates the study of protein disorder regions and provides an important reference for further study of the functions of these regions.


For more details of the competition, please refer to:

https://www.nature.com/articles/s41592-021-01123-5

https://www.nature.com/articles/s41592-021-01117-3


(Translated by Chengcheng Chang, edited by Daniel Stefan and JianjingYun)