TumorScope - D2Deep
Developed by Konstantina Tzavella with the collaboration of the Bio2Byte research group (VUB, Brussels, Belgium)
This is a web server that provides quick predictions for the effect of mutations on proteins and visualization tools for a comprehensive view of mutational landscape
Explore the predictions for a particular protein by using the Menu Proteins or Advanced search above.
Abstract
Undoubtedly, the accurate assessment of the pathogenicity of genetic variants associated with human diseases holds immense significance. It can contribute to unravelling the complex biological mechanisms and shape clinical decisions in a profound manner. However, it remains concerning that a staggering majority (over 98%) of these variants still harbor unknown consequences [1-3].
We have developed a novel approach to predict the effect of the mutations in the proteins. The method capitalizes on the abundance of available protein sequences, state-of-the-art Large Language Model architectures and Evolutionary Information from Multiple Sequence Alignments. The statistical model incorporated makes possible the calculation of confidence score for each prediction facilitating its interpretation.
We have named this method D2Deep, and you can easily use it on this web server. The tool only needs as inputs the protein/gene name and, optionally, the mutation itself to predict the results. We hope that this new tool will be useful to assess the effect of mutations and shed more light on the complex cellular processes.
Publications and public data
Combining evolution and protein language models for an interpretable cancer driver mutation prediction with D2Deep[4]
Konstantina Tzavella
,
Adrián Díaz,
Catharina Olsen,
Wim Vranken
Read the full publication in the journal Briefings in Bioinformatics or download datasets with predictions, training data, epistatic features, and more for all proteins modeled to date in Zenodo.
Features
Our web server provides a comprehensive set of tools for querying and visualizing protein mutations based on D2Deep data.
With its user-friendly interface and powerful visualization capabilities,
this server makes it easy for researchers and clinicians to explore the impact of protein mutations and
develop new insights into the underlying biology.
Results tables can be downloaded in CSV format as well as the biophysical plots in PNG format.
Usability
Our web server is designed to be easy to use, even to those without an extensive background in bioinformatics.
The interface is user-friendly and intuitive, and the visualizations are designed to be interactive and informative.
Once a mutation is selected, the server generates a report that includes the biophysical information about the mutation, the impact it has on the protein and the confidence score of the prediction.
Terms of Service
This website may use essential cookies to ensure its proper operation, as well as Google Analytics tracking to understand how you interact with it.
Google Analytics helps us improve the website by collecting and reporting information on its usage.
Citations
- Van Hout, C. V. et al. Exome sequencing and characterization of 49,960 individuals in the UK Biobank. Nature 586, 749-756 (2020)
- Karczewski, K. J. et al. The mutational constraint spectrum quantified from variation in 141,456 humans. Nature 581, 434-443 (2020)
- Landrum, M. J. & Kattman, B. L. ClinVar at five years: delivering on the promise. Hum. Mutat. 39, 1623-1630 (2018)
- Tzavella, K. & Díaz, A. & Olsen, C. & Vranken, W. Combining evolution and protein language models for an interpretable cancer driver mutation prediction with D2Deep. Briefings in Bioinformatics 26, Issue 1 (2025)