yanismiraoui / M4R-dash

WEB APP : Deep unsupervised learning methods for the identification and characterization of TCR specificity 🫁🧬

This dashboard aims to show how perform the different deep learning methods described in the paper Deep unsupervised learning methods for the identification and characterization of TCR specificity, when given a real CDR3 sequence or a personnalized one.

Guidelines:

1. Choose a model from which the representation will be computed.
2. Type your own CDR3 sequence or click on the 'Generate' button to generate an existing CDR3 sequence at random.
3. You can also choose random or personnalized v-gene and j-gene.
4. It is also possible to compare all the models together on the same TCR in the Compare models tab.
5. If you have any questions about this research, a chatbot is also ready to answer any questions that you may have in the ChatBot tab.

Results:

• The predicted embedding is computed and displayed.
• The predicted clustering is also displayed below the plot.

PLEASE NOTE: the website can sometimes be slow to load the text and the predictions. Please wait a few seconds for the content to load. This web application is hosted on Replit and Heroku.

🔗 Paper

🔗 Website of the demo

🔗 Github repository of the main analysis