P3CMQA: Single-Model Quality Assessment Using 3DCNN with Profile-Based Features

P3CMQA: Single-Model Quality Assessment Using 3DCNN with Profile-Based Features Yuma Takei http://orcid.org/0000-0001-7442-0022 Takashi Ishida http://orcid.org/0000-0002-9478-3223

Model quality assessment (MQA), which selects near-native structures from structure models, is an important process in protein tertiary structure prediction. The three-dimensional convolution neural network (3DCNN) was applied to the task, but the performance was comparable to existing methods because it used only atom-type features as the input. Thus, we added sequence profile-based features, which are also used in other methods, to improve the performance. We developed a single-model MQA method for protein structures based on 3DCNN using sequence profile-based features, namely, P3CMQA. Performance evaluation using a CASP13 dataset showed that profile-based features improved the assessment performance, and the proposed method was better than currently available single-model MQA methods, including the previous 3DCNN-based method. We also implemented a web-interface of the method to make it more user-friendly.
03 19 2021 40 bioengineering8030040 Japan Society for the Promotion of Science http://dx.doi.org/10.13039/501100001691 18K11524 https://creativecommons.org/licenses/by/4.0/ 10.3390/bioengineering8030040 https://www.mdpi.com/2306-5354/8/3/40 https://www.mdpi.com/2306-5354/8/3/40/pdf 10.1002/prot.25834 10.1002/prot.25823 10.1002/prot.25697 10.1093/bioinformatics/bty494 10.1371/journal.pone.0221347 10.1186/1471-2105-13-224 10.1038/srep33509 10.1093/nar/25.17.3389 10.1093/bioinformatics/btu352 Ioffe Batch Normalization: Accelerating Deep Network Training by Reducing Internal Covariate Shift Proceedings of the 32nd International Conference on Machine Learning, Lille, France, 7–9 July 2015 2015 Volume 37 448 10.1109/ICCV.2015.123 10.1109/CVPR.2016.90 10.1093/bioinformatics/btt473 10.1093/nar/gkg571 10.1002/prot.21767 10.1002/prot.22589 10.1002/prot.23200 10.1002/prot.24452 10.1002/prot.25415 10.1002/prot.22488 10.1093/bioinformatics/btw819 10.1093/bioinformatics/bty1037 10.1002/prot.25278 RMSprop loses to SMORMS3—9Beware the Epsilon! 2015 https://sifter.org/~simon/journal/20150420.html 10.1093/bioinformatics/bty419