Girinath G. Pillai

Gromacs is a popular, free and Open Source software suite for high-performance Molecular Dynamics (MD) and output analysis. MD is a computer simulation technique used to study the physical movements of atoms and molecules over time, which has plenty of applications in Material Science, Drug Discovery and Chemical Reaction Modelling. Although Gromacs is a very powerful tool, it requires a lot of manual work to prepare an end-to-end simulation and analysis pipeline. I have started a GitHub repo (link in the comments), where I am building a local Agent with the intention to move away all the annoying work from the scientists that use Gromacs, so that they could focus on their main tasks, where they can apply in full their domain knowledge and expertise. This work follows few main principles: - Everything stays local (where local could be a scientist laptop, or a local server). Nothing must leave the user's organisation network. - Open Source components only. - Small Language Models only. - The target User Experience is based on the BYOP (Bring Your Own PDB) and MD expertise principle :) I have chosen the Hugging Face's Smolagents lightweight framework for this and the local model that to date is providing the best performance is Qwen 2.5 3B Instruct, in FP 16 (but that wouldn't necessarily be the final choice). Running an AI Agent locally and using only Open Source components gives you more control, but still some hurdles can be met, so always think about robust fallback strategies. Please be patient, as this is still a work in progress (I have started it this week only, following a conversation with few Computational Chemistry practitioners), but it also serves as a playground to apply some of the techniques I have explained in my latest book https://mng.bz/z2jA #python #smolagents #SLM #opensource #moleculardynamics #gromacs #generativeai #chemistry Manning Publications Co. Hugging Face

65

5 Comments

Ferrous ascorbate, BA of which is around 67% (according to a ref.), is well known for treatment of IDA, but here is another story of Ferrous Ascorbate! Actual Study-- A group of researchers wanted to understand----Can the tiny drops of fluid inside your eyes reveal if your diabetes is quietly damaging your vision? So, they studied 125 patients — some with diabetic retinopathy and some without. They looked not just at what doctors usually check, but went deeper ..... down to the chemical level inside the eye. Using cutting-edge technology and ML, they discovered that: #The eye fluid of people with diabetic retinopathy had 20 key chemical differences. #These differences were tied to how the body manages energy, amino acids, and nutrients. One of the most surprising observations was that.......#Ferrous ascorbate showed a strong link to poor blood sugar control. These changes weren’t random , they told a story of disrupted inner chemistry, giving clues to why diabetic eye disease happens and how it could possibly be detected much earlier. Because catching diabetic retinopathy early can mean the difference between clear vision and vision loss. And of course, if something as simple as a chemical pattern in the eye can be used like a fingerprint to detect risk , it opens the door to personalized, preventive care. Study : Fe Ascorbate as a potential biomarker for diabetic retinopathy

10

Just saw an interesting paper in JMC - a comprehensive review of AI-powered drug discovery. This is not how I would have written the paper as it kind of mixes the pre-deep learning legacy approaches that later got rebranded as AI with modern and frontier approaches but it is a good review everyone new to the field should read. For the next semester you can teach it in the classroom. And if you are a medicinal chemist, highlight the repurposed and slightly modified structures. https://lnkd.in/g3uet8vy

118

1 Comment