AI BioInformatics w/ multiple Agents

AI research explores three AI Agents in parallel to explore medical substance repurposing for new medicines.
DTI (drug-target-interaction) affinity as calculated by a biomolecular trained AI agent, by a knowledge graph agent (knowledge from databases) and by a search agent (scientific publications).

This AI research demonstrates the potential of AI to revolutionize medical substance discovery by automating and accelerating the process of identifying and validating potential medical substance candidates. This approach utilizes a multi-agent system, where each agent performs a specific task related to information retrieval, analysis, and prediction, ultimately providing a comprehensive assessment of potential DTI.

The first agent, a machine learning model trained on a binding database, predicts the likelihood of binding between a medical substance and its target protein. This model combines message passing and convolutional neural networks to effectively capture the structural and spatial features of both the medical substance and its target.

The second agent, a knowledge graph agent, leverages multiple biomedical databases to construct a knowledge graph. By analyzing the shortest path between the medical substance and target protein within this graph, this agent provides valuable insights into the existing research and potential interactions between the two.

Finally, a search agent utilizes text analysis to extract relevant information from scientific literature, further validating the predicted binding affinity. The combined results from these three agents offer a comprehensive assessment of potential DTI, enabling researchers to prioritize and validate promising candidates for further investigation.

This AI-driven approach has the potential to accelerate the medical substance discovery process, enabling faster development of new treatments and potentially reducing the cost of medical substance development.

All rights with Authors:
DRUGAGENT: EXPLAINABLE DRUG REPURPOSING AGENT WITH LARGE LANGUAGE MODEL-BASED REASONING

Genesis: Towards the Automation of Systems Biology Research

00:00 AI Bioinformatics Start
02:26 ML Agent 1 Molecular and Protein Encoders
06:42 MPNN_CNN_BindingBD Model
08:30 DeepPurpose (Harvard Univ, Georgia Tech)
09:08 Knowledge Graph Agent 2
11:30 Search Agent 3 on scientific papers
14:34 Pre-print “DrugAgent: Explainable Drug Repurposing Agent”
19:19 CODE DrugAgent (GitHub)
21:22 Pre-print Genesis: Automation of System Biology Research

#airesearch
#science
#medicine

source