Loading...

Loading
DeepCLIP is a novel deep learning tool for finding binding-preferences of RNA-binding proteins.

Summary of DeepCLIP and its functionalities:
DeepCLIP is a neural network with shallow convolutional layers connected to a bidirectional LSTM layer.
DeepCLIP outputs binding profiles and overall scores for each input sequences.
Binding profiles show whether areas of sequences contain possible binding sites or whether they look like random genomic background.

On this webserver you can use our pretrained models to predict on your own sequences, or train your own model.
However, for optimal training, we highly recommend using CV runmode on your own server with a local DeepCLIP installation.
You can navigate using the menu at the top. New users are encouraged to read the DeepCLIP guide first.

About

DeepCLIP was developed by Alexander GB Grønning and Thomas K Doktor, with additional code by Simon J Larsen.

Citation

If you use DeepCLIP in your work, we kindly ask you to cite the following paper:
Grønning AGB & Doktor TK, et al. DeepCLIP: predicting the effect of mutations on protein-RNA binding with deep learning. Nucleic Acids Res. 2020;48(13):7099-7118.
doi:10.1093/nar/gkaa530

Disclaimer

While we provide pre-trained models and the ability to train your own, we take no responsibility for the results produced by this webserver.
Results are provided as is and with no guarantee of truth or validity.
Data uploaded to the server is stored securely and linked to a unique and anonymous job id. Anyone in possession of the job id may access the data uploaded.
Data on the server may be deleted without further notice and you should not rely on availability of models, including both pre-trained and user-trained models.

Multi-prediction

Sequences
Use example sequences
Predict paired sequences
Use example sequences
Enable long sequence prediction mode.

Predictions

Plot profile difference (with variants only)
Download profile plots Download profile data

Training model

DeepCLIP is currently training your model. This page will automatically redirect once the training is complete.

You can bookmark this page and come back later.

Model summary

Foreground sequences Background sequences Download model weights and all summary plots

ROC curve

CNN filters ranked by score

Prediction score distribution

Sequences
Use example sequences
Predict paired sequences
Use example sequences
Enable long sequence prediction mode.

Predictions

Plot profile difference (with variants only)
Download profile plots Download profile data
Model training failed

Log


        

Binding sequences

Select how binding sequences will be provided. If BED file is selected, a target species and assembly must be selected as well.

Background sequences

Select how background sequences should be generated.

  • If FASTA file is selected a FASTA file containing sequences must be provided.
  • If Generate from BED file is selected sequences will be generated from the provided BED file containing binding sequences.
  • If Shuffle input is selected sequences will be generated by reshuffling the provided binding sequences.

Sequence preprocessing

  • Min. sequence length: all sequences shorter than this will be discarded.
  • Max. sequence length: all sequences longer than this will be discarded.
  • BED file only: If Pad sequences is given all sequences will be padded with this number of surrounding bases.
  • BED file only: If Fixed sequence width is given all sequences will be trimmed/padded to fit this length.

Data split

Specify how sequences should be split for training, validation and testing.

Program control


deepclip-shiny version 0.3