DeepFRI

Gligorijević et al. |

DeepFRI

Omics

Run
About
API Example

Step 1: Upload your data

Upload .FASTA/.PDB/.NPZ File

Drag your file(s) or upload

Your file can be in the following formats:fasta, npz, pdb

Three file formats are acceptable as input. For the FASTA file, multiple amino acid sequences can be specified. You can see an example structure below. >protein1 AKSMDTTDIGAF… >protein KPVTLYDIAGFFA… For contact maps prediction, file format should be in .npz format. The .npz file format is a compressed collection of files with variable-based names. In the .npz compressed file, there should be .npy files in the example. For example, there are three .npy files inside the .npz file (C_alpha.npy, C_beta.npy and seqres.npy). For more detail please check the method section of the article. On the other hand, a text-based file format called Protein Data Bank (PDB) is used to store information about the three-dimensional molecular structures. If user wants to make a prediction from three-dimensional molecular structures, .pdb format should be provided.

Don’t have a file?

Use our demo data to run

Use Demo Data

View example data

Step 2: Set Parameters

Ontology Type

GO-Molecular Function

Step 3: Complete run profile

Job name - Optional

One of the most significant biological challenges in the post-genomic era is understanding the functional roles and investigating the mechanisms of newly found proteins. A deep learning technique called DeepFRI uses both sequences and contact map representations of 3D structures to predict protein function. The protein structures from the PDB and SWISS-MODEL are used to train DeepFRI. LSTM-LM(Long Short-Term Memory Language Model) was used to learn features from protein sequences and GCN was used to discover features from contact maps.

Example use case: Predicting protein functions

Technology: Graph Convolutional Networks (GCN), LSTM-LM

Limitation: Some of the options to predict protein functions are currently not available. Please check this page for more information.

Metrics: Some of the metrics related to work can be found here

Citation:

Gligorijević, V., Renfrew, P.D., Kosciolek, T. et al. Structure-based protein function prediction using graph convolutional networks. Nat Commun 12, 3168 (2021). https://doi.org/10.1038/s41467-021-23303-9

Released:
Dec-13-2022

Previous Job Parameters

Your previous job parameters will show up here
so you can keep track of your jobs

Results

Parameters