Fine-tuning GPT-3 for machine learning electronic and functional properties of organic molecules.

We evaluate the effectiveness of fine-tuning GPT-3 for the prediction of electronic and functional properties of organic molecules. Our findings show that fine-tuned GPT-3 can successfully identify and distinguish between chemically meaningful patterns, and discern subtle differences among them, exhibiting robust predictive performance for the prediction of molecular properties. We focus on assessing the fine-tuned models' resilience to information loss, resulting from the absence of atoms or chemical groups, and to noise that we introduce via random alterations in atomic identities. We discuss the challenges and limitations inherent to the use of GPT-3 in molecular machine-learning tasks and suggest potential directions for future research and improvements to address these issues.

Network pharmacology and molecular docking-based study on exploring the potential mechanism of Lycium barbarum L: In the treatment of atherosclerosis.

BACKGROUND: Goji berries (Lycium barbarum L) are herbal medicine that have a long history of use and multiple pharmacological activities. In this study, we investigated the potential therapeutic effects of Goji berries on atherosclerosis (AS) using network pharmacology and molecular docking. METHODS: The active compounds of Goji berries were identified using the Traditional Chinese Medicine Systems Pharmacology platform, as well as the literature and the targets of each active compound were obtained using the Swiss Target Prediction database. The AS-related targets were collected from the GeneCards and OMIM databases to obtain the common targets of Goji berries and AS. The drug-compound-target-disease network and protein-protein interaction network were constructed using the Cytoscape software to obtain the core target proteins of Goji berries related to AS. Gene ontology analysis of the core targets and Kyoto encyclopedia of genes and genomes pathway enrichment analysis were performed by Metascape. The target-chemical correlations were verified using AutoDock molecular docking. RESULTS: After analysis, 44 active compounds within Goji berries were obtained that exhibit associations with AS. Among these, the proteins exhibiting the highest degrees of interaction within the compound-targeted protein protein-protein interaction network were AKT1, SRC, MAPK3, MAPK1, RELA, and STAT3. The gene ontology-biology process analysis showed that compound-targeted proteins were mainly involved in regulating small molecule metabolic process, cellular response to chemical stress, reactive oxygen species metabolic process, and regulation of inflammatory response. Kyoto encyclopedia of genes and genomes pathway mainly included lipid and AS in which AKT1, SRC, MAPK3, and MAPK1 were involved. Advanced glycation end-product-receptor for advanced glycation end-product signaling pathway in diabetic complications, Chagas disease, and pancreatic disease. Molecular docking assessment showed that fucosterol is bound to AKT1, MAPK3, and SRC. CONCLUSION: This study demonstrates that network pharmacology and molecular docking analyses contribute to a better understanding of Goji berries active compounds and targets as potential therapeutic drugs for treating AS.