Accurate De Novo Prediction of Protein Contact Map by Ultra-Deep Learning Model.
Citations
10,601 citations
4,833 citations
2,097 citations
1,491 citations
1,026 citations
Cites methods from "Accurate De Novo Prediction of Prot..."
...We evaluated our extensions to previous approaches by generating a baseline model to predict distances only, with no MSA subsampling and selection; the contact prediction accuracy of this network is comparable to previously described models (3, 12, 19, 20)....
[...]
...The overall architecture of the network is similar to those recently described for distance and contact prediction (3, 4, 7, 12)....
[...]
...%) of the top L predicted contacts on CASP13 and CAMEO targets Method CASP13 FM domains CAMEO very hard targets s ≥ 24 s ≥ 12 s ≥ 24 s ≥ 12 RaptorX-Contact 44.7 61.3 NA NA TripleRes 42.3 60.9 NA NA trRosetta 51.9 70.2 48.0 62.8 Baseline* 44.3 60.7 41.6 57.5 Baseline+1† 46.0 62.2 43.1 57.4 Baseline+1+2‡ 48.2 64.6 44.4 58.7 Baseline+1+2+3§ 51.3 69.3 46.1 61.4 The values for other methods are slightly different from those listed on the CASP13 website (http://predictioncenter.org/casp13/), probably due to different treatment of target length L (i.e., length of full sequence or length of domain structures; the latter is used here)....
[...]
...To compare our Rosetta minimization protocol (trRosetta) to CNS (8), we obtained predicted distance restraints and structure models for all CASP13 FM targets from the RaptorX-Contacts server, which uses CNS for structure modeling (4), and used the distance restraints to generate models with trRosetta....
[...]
...Following RaptorX-Contact (4, 12) and AlphaFold (7), we learn probability distributions over distances and extend this to orientation features....
[...]
References
123,388 citations
73,978 citations
[...]
46,982 citations
40,257 citations
30,843 citations
"Accurate De Novo Prediction of Prot..." refers methods in this paper
...To speed up training, we also add a batch normalization layer (Ioffe and Szegedy, 2015) before each activation layer, which normalizes its input to have mean 0 and standard deviation 1....
[...]