Distance-based Protein Folding Powered by Deep Learning


This web server predicts local struture properties, contact and distance matrix, inter-residue orientation and tertiary structure of a protein using an ultra-deep convolutional residual neural network from primary sequence or a multiple sequence alignment. It works particularly well on proteins without many sequence homologs. This server was ranked 1st in contact prediction in both CASP12 and CASP13. See here for an official ranking list published by the CASP12 assessors and 1, 2 for the method description. Currently the server will predict contact/distance matrix for a protein of no more than 1300 amino acids, and 3D models for a protein of no more than 1000 amino acids.


Job Identification
: :

Sequence for Prediction      (example)
:                 check here if the input is a single Multiple Sequence Alignment (notes and format)

: check here if only contact/distance matrices are needed

Current server load
1374 jobs pending
634 jobs done in the last 24 hours
6919 jobs done in the last 30 days

#server users: 66434
#processed jobs: 544566

Job policy
To maximize the utility of the server to the community at large the following limits on job submission for each user are enforced.
  • Please submit up to 10 protein sequences (in the FASTA format) in a batch for prediction. If your email box cannot receive a large attachment file, please submit a SMALL number of sequences in a batch. Otherwise the result package may be too big to be received.
  • Please SAVE the assigned JobID for retrieval of job results, especially when an email address is not provided in submission.
  • Each user can have no more than 500 sequences pending prediction at any time.
  • Should you have a special project that requires more resources, please contact us to inquire further.