RoseTTAFold 

Source-This post on RoseTTAFold has been created based on the article “The use of AI in drug development | Explained” published in “The Hindu” on 20 May 2024.

Why in News?

The Introduction of Artificial Intelligence (AI) tools like RoseTTAFold has greatly accelerated the process of drug development offering new possibilities for rapid advancements.

About RoseTTAFold 

1. About: It is a software tool that uses deep learning to quickly and accurately predict protein structures based on limited information.

The software is built on a “three-track” neural network that analyzes protein sequences, interactions between amino acids, and potential three-dimensional structures simultaneously.

2. Developed by: It was developed by researchers at the University of Washington, U.S.

3.Technology and Function

i) AI Model: It utilizes generative diffusion-based architectures which is a type of AI model to predict structural complexes.

ii) Deep Learning: It employs deep learning techniques to quickly and accurately predict protein structures from limited data.

4. Capabilities

i) Protein Structure Prediction: It is capable of predicting not only static protein structures but also protein-protein interactions.

ii) Versatile Predictions: It can predict interactions and structures involving combinations of protein, DNA, and RNA.

iii) Efficiency in Structural Determination: The tool significantly reduces the time required to determine protein structures. This process can otherwise take years of laboratory work using traditional methods.

5. Drawback:

i) Accuracy Limitations: AI tools like RoseTTAFold generally offer up to 80% accuracy in protein interaction predictions, with significantly lower accuracy for protein-RNA interactions.

ii) Scope of Application: These tools primarily aid in the target discovery phase of drug development and do not contribute to later stages such as pre-clinical and clinical trials, limiting their impact on the overall success of drug development.

iii) Model Hallucinations: The diffusion-based architectures can suffer from “model hallucinations,” producing inaccurate or nonexistent predictions when training data is insufficient.

iv) Restricted Access: Newer versions, such as AlphaFold 3, have not released their source code, limiting independent verification and broader scientific use, particularly in protein-small molecule interaction studies.

Challenges and Opportunities for AI in Drug Development in India

1. India faces a significant gap in the availability of skilled AI scientists compared to countries like the U.S. and China. This has hindered India’s ability to secure a first-mover advantage in

2. Despite India’s strong background in protein X-ray crystallography, modeling, and structural biology, the lack of expertise in AI has been a limiting factor.

3.  India has a rapidly expanding pharmaceutical sector, providing a substantial opportunity for the country to excel in applying AI tools in target discovery, identification, and drug testing.

UPSC Syllabus: Science and technology (Health)