Deep Learning for Predicting Protein Interactions
Author Information
Author(s): Majila Kartik, Ullanat Varun, Viswanath Shruthi
Primary Institution: National Center for Biological Sciences, Tata Institute of Fundamental Research, Bangalore, India
Hypothesis
Can a deep learning method accurately predict interactions for intrinsically disordered regions of proteins?
Conclusion
The Disobind method outperforms existing models in predicting inter-protein contact maps and interface residues for intrinsically disordered proteins.
Supporting Evidence
- Disobind predicts inter-protein contact maps and interface residues for IDRs and their partners.
- Disobind outperforms AlphaFold-multimer and AlphaFold3 across multiple confidence cutoffs.
- Combining Disobind with AlphaFold predictions improves performance further.
Takeaway
Scientists created a computer program that helps predict how certain proteins interact with each other, especially when they are not structured well.
Methodology
Disobind uses a deep learning model trained on sequence data to predict inter-protein contact maps and interface residues.
Potential Biases
The model may be biased due to the limited number of experimental structures available for IDRs.
Limitations
The method is limited to binary IDR-partner complexes and requires both proteins to be known to bind.
Digital Object Identifier (DOI)
Want to read the original?
Access the complete publication on the publisher's website