logo

RocketX is a de novo protein structure prediction algorithm by incremental inter-residue geometries prediction and model quality assessment using deep learning. In RocketX, a closed-loop feedback mechanism is constructed by inter-residue geometries prediction (GeomNet), structure simulation, and model quality assessment (EmaNet) to progressively improve the accuracy of predicted model. The structural model folded based on the geometric constraint predicted by GeomNet is evaluated by EmaNet, and the inter-residue distance deviation and per-residue lDDT estimated by EmaNet are fed back to GeomNet as the dynamic features to correct the geometric constraints prediction and progressively improve the structure model. The experimental results on the benchmark testset, CASP14 FM targets and CAMEO targets show that the performance of RocketX is comparable to the state-of-the-art server methods.


  Rocket On-line Server [View example of output]

Please input the protein sequence (mandatory, recommended ≤ 800 amino acids, Click for an example input) below.
Or upload the protein squence file:
Email: (mandatory, where results will be sent to)

Job name: (optional, your given name to this job,numbers and letters)


RocketX news

  • The 1-week CAMEO blind test (2022-01-15) of Rocket is released.
  • The 1-week CAMEO blind test (2022-01-08) of Rocket is released.
  • The 1-week CAMEO blind test (2022-01-01) of Rocket is released.
  • The 1-week CAMEO blind test (2021-12-25) of Rocket is released.
  • The 1-week CAMEO blind test (2021-12-18) of Rocket is released.
  • The 1-week CAMEO blind test (2021-12-11) of Rocket is released.
  • The 1-week CAMEO blind test (2021-12-04) of Rocket is released.

References

  • Jun Liu, Guang-Xing He, Kai-Long Zhao and Gui-Jun Zhang*. De novo protein structure prediction by incremental inter-residue geometries prediction and model quality assessment using deep learning. bioRxiv, doi: 10.1101/2022.01.11.475831.